Published On: 2/20/2025

Causal Inference for Regulatory-Grade Evidence Generation: Behind the Data

Interview with David Pritchard, PhD, Target RWE Director of Data Management and Statistics

Real-world data (RWD) is revolutionizing how pharmaceutical companies develop and bring novel therapies to market. However, extracting high-quality insights from RWD requires sophisticated epidemiological algorithms and innovative statistical methods to address the inherent complexities of data. Enter causalStudio™ - Target RWE's cutting-edge analytical platform with two integrated offerings to enhance data-driven decision-making: causalRisk™ and causalPHR™. causalRisk™ is a simplified solution to the complexity of estimating causal effects of outcomes that seamlessly connects with causalPHR - a dynamic visual platform to present epidemiological study findings in a clear and compelling format for regulatory-grade evidence generation.

Our interview with Target RWE Director of Data Management & Statistics, David (Dave) Pritchard, PhD, delves into the core capabilities of causalStudio™, focusing on its advanced statistical methods, robust data handling, and ability to generate impactful visualizations. Learn how causalStudio™ addresses key challenges in RWD analysis, including bias mitigation and real-world evidence suitable for regulatory submission, as well as user-friendliness and potential to transform how researchers leverage RWD to inform drug development and improve patient outcomes.

How does causalPHR™ complement the causalRisk™ analytical solution within causalStudio™?

Dave: causalRisk™ and causalPHR™ are ideal complements of each other within the causalStudio™ platform. causalRisk™ allows analysts to apply principled statistical methods to complex epidemiological study designs to gather valuable clinical insights. On the other hand, causalPHR™ provides a sophisticated data visualization platform for presenting study results to different audiences, that may include internal research partners, decision makers, or the broader public. causalPHR™ supports the easy creation of crucial information to support epidemiological studies such as descriptive tables of study populations, cumulative incidence and survival curves for safety or adverse events, and Sankey and sunburst figures for describing treatment patterns - among many other types of visual displays. Additional valuable features for supporting project workflows include integrated commenting among collaborators, project versioning facilities, and selective access controls for partners.

What specific validated analytics are used in causalStudio™ and how have they been proven and trusted for regulatory decision-making?

Dave: causalRisk™ provides a collection of validated statistical routines that estimate the cumulative risk of a right-censored counterfactual outcome that may be subject to dependent right censoring, confounding, selection bias, and competing risks. These routines are verified by a suite of automated tests that compare the results produced by the package to those calculated by creating special cases of the estimators for a given situation using only standard R packages. causalRisk™ has been used to support submissions to various regulatory agencies including the FDA, Chinese FDA, and Health Canada.

Can you provide examples of the advanced epidemiological algorithms and innovative statistical methods used in causalStudio™?

Dave: Some advanced statistical methods provided by causalRisk™ include cumulative risk inverse probability weighted estimators (including doubly-robust estimators), cumulative risk g-formula estimators, hazard ratio estimators, and cumulative count inverse probability weighted estimators.

How does causalStudio™ address the challenge of bias in real-world data analysis? What specific methods are used to account for confounding variables? 

Dave: Confounding variables can cause bias if not properly accounted for; causalRisk™ offers researchers methods that can address potential sources of bias that often arise in observational studies, such as differential treatment propensity and loss to follow-up based on a patient’s health status. The methods can be used to model mechanisms of interest and if models are correct, then researchers can recover an unbiased estimate of the target quantity.

Causal inference is critical for generating regulatory-grade evidence from real-world data, and Target RWE’s causalStudio™ provides researchers with the advanced analytics needed to meet this challenge. By integrating validated statistical methods with a dynamic visualization platform, causalStudio™ enables precise estimation of causal effects, robust bias mitigation, and clear presentation of epidemiological insights.

To learn more about how causalStudio™ can support your research, contact info@targetrwe.com today and stay tuned for more articles on our full suite of analytical capabilities!