Our summary
Clinical trials are pivotal for assessing treatment effects, employing defined protocols to minimize confounding. Challenges persist in managing real-world data, including confounding and missing data from events like treatment discontinuation or patient loss to follow-up.
The Estimand framework (EF) addresses these challenges, defining treatment effects and ensuring consistency in study design and analyses. Simultaneously, the Target Trial framework (TTF) minimizes biases in observational studies by emulating a hypothetical trial protocol with observational data.
In medical research, Causal inference frameworks like the EF and TTF refine scientific questions. They complement each other, allowing stakeholders to align on comparative effectiveness research questions in a principled manner, reducing ambiguity and biases in observational studies.
In this study, researchers from Roche/Genentech and Flatiron Health illustrate the use of EF and TTF in comparing overall survival for people with advanced NSCLC treated with front-line chemotherapy (within a clinical trial vs routine clinical-care).
Why this matters
This study is important because it shows that combining the EF and TTF approaches can enhance the quality of design and analysis in comparative effectiveness studies that involve real-world/observational data (RWD). Without careful design, RWD comparative effectiveness studies are prone to ill-defined questions and errors that can be random or systematic. While increasing the study sample size can address random errors, rigorous study design and statistical approaches may address systematic errors, but an ill-defined question reduces the overall usefulness of the study. The EF and the TTF can help mitigate common study pitfalls by guiding researchers to define the research question transparently, state the corresponding comparative measure (causal contrast), assess the estimability of the causal contrast given the available data, and articulate any assumptions.