Presentation (CTI)

OBJECTIVES: Sex and age are important determinants of disease risk and treatment response, yet their systematic consideration in real-world evidence (RWE) remains inconsistent. Sex- and age-stratified analyses are increasingly recommended in RWE, post-authorization safety studies (PASS) and outcome research, but are often conducted selectively. We developed a practical framework to standardize the specification, implementation, and reporting of sex- and age-stratification to support balanced interpretation of both the presence and absence of subgroup differences.
METHODS: We developed a stratification framework that can be applied to existing RWE pipelines. The framework requires a priori definitions for sex and age categories, and it standardizes a core set of outputs such as stratified baseline characteristics, subgroup-specific effect estimates, diagnostics on sparsity and estimator stability and a reporting layout. Feasibility will be demonstrated by re-analyzing a completed study to introduce prespecified sex- and age-stratified outputs.
RESULTS: The framework generated a consistent evidence package that can be applied across therapeutic areas and study designs without changing the underlying epidemiologic approach. A toolkit accompanies the framework, including a lightweight programming template for repeatable subgroup estimation and tabulation, a reporting structure that makes both scenarios transparent, irrespective of the observed differences, considerations for interpretation and decision rules for when to present subgroup-specific estimates versus pooled estimates only, based on sample size, event counts, and estimate stability.
CONCLUSIONS: A stratification framework can improve efficiency, reproducibility, and credibility, and offers an opportunity to optimize product positioning in specific sub-groups by having sex and age stratification as a default and by treating null heterogeneity as meaningful evidence when communicated responsibly. Without this, the risk of uncritically generalizing pooled findings to women and age-specific subgroups persists, potentially contributing to treatment guidelines and product-level evidence that fail to reflect the full clinical picture across the populations they are intended to serve.