2. September 2025

Annual conference of the International Society for Clinical Biostatistics (ISCB) in Basel

Statisticians from academia, industry, and regulatory bodies gathered to exchange insights on the latest developments in theoretical approaches and practical tools across the fields of clinical trials, real-world data, and epidemiology.

As an international professional society for clinical biostatistics, ISCB is dedicated to the development of methods that are relevant to medical research. This year, one of the recurring themes was the improved use of routinely collected clinical data. Such data can be purposefully integrated into studies to enhance statistical efficiency – particularly important for smaller studies, such as those investigating rare diseases. Another topic was how study centres should be accounted for in the analysis, as populations and procedures may vary between centres. Solutions were discussed on how to account for these differences during the planning of analyses in order to strengthen the validity and interpretability of study results.

Data quality and variable handling
Speakers highlighted the risks associated with assuming simple linear relationships between continuous variables such as age or weight and study outcomes. The still common practice of dichotomisation was also critically examined. This refers to dividing continuous or ordered variables (such as a mRS scale from 0 to 6) into two categories. Dichotomisation leads to a loss of valuable information, unnecessarily large sample size and prolonged study duration. Easily implemented visual aids were presented to help assess the assumptions made about variables.

Identification of causal inference
Estimands are precise specifications of the treatment effect of interest that a study aims to determine. They are defined by the target population, the treatments to be compared, the endpoint (including the time of measurement), and a strategy for handling events occurring during the study (intercurrent events). Estimands must be clearly defined in advance, as they guide the study design. For a robust definition of estimands and the identification of potential events, discussion between statistical experts and clinical professionals is essential — this was the general consensus at the conference.

The DKF statistics team is available to support clinical researchers in exactly this kind of collaboration. The earlier their experience and expertise are incorporated into the planning of a study, the more targeted and efficient the study can be designed, conducted, and analysed.