Lecture in the Bremen Colloquium Epidemiology - Public Health: Using Causal Inference to Leverage Nationwide Healthcare Data for Breast Cancer Research

The lecture will take place from 11:30 a.m. in room 1550 and as a Zoom meeting. The lecture

is in English language.

uni-bremen.zoom-x.de/j/69772773883

Meeting-ID: 697 7277 3883

Kenncode: 383784

Information about the lecture

Large medical datasets sourced from electronic health records and administrative claims are increasingly available for research in several countries, including France. These datasets are of great value for studying rare outcomes, measuring the effects of interventions when clinical trials are not available, or extending the applicability of clinical trial results to broader, more representative populations. However, association methods traditionally used for observational studies are susceptible to numerous biases that may prevent causal interpretation of their results. Alternatively, causal inference methods, including target trial emulation, aim to frame research questions in terms of causal estimation, thereby circumventing biases associated with flawed study designs and lack of randomization. In this context, the French Early Breast Cancer Cohort (FRESH) - encompassing data from over 235,000 patients diagnosed with early breast cancer, derived from the French national social security system - provides a unique research opportunity. This talk will delve into this comprehensive dataset and discuss findings from two causal inference analyses we conducted. The first analysis evaluates the oncologic safety of vaginal estrogen therapy post-early-stage breast cancer diagnosis. The second analysis explores the intricate relationships among young age at diagnosis, adherence to endocrine therapy, and breast cancer outcomes.