IDDI speaking at the 9th Statistics & Biopharmacy Conference (SnB2022)

September 19, 2022 — September 21, 2022

Paris, France

Join IDDI team at the 9th Statistics & Biopharmacy Conference (#SnB2022) by the French Society of Statistics (SFdS)


  • IDDI’s Chief Scientific Officer Marc Buyse, ScD invited lecture on: Lessons Learnt for Clinical Research from the COVID-19″
  • Mickaël De Backer, Senior Research Biostatistician, IDDI’s BENEFIT team invited speaker: “The use of Generalized Pairwise Comparisons for designing trials that are tailored to patients’ wishes and needs”
Mickaël De Backer, Senior Research Biostatistician, IDDI


The standard design of clinical trials is based on a primary outcome that is considered clinically relevant and statistically sensitive for detecting a difference between experimental treatment and standard of care. Although treatment may affect secondary outcomes, these are typically ignored inferentially at the design and analysis stages. Consequently, trials typically make suboptimal use of information, and leave very limited room to the voice of patients for expressing the impact of several treatment dimensions.

The method of Generalized Pairwise Comparisons (GPC) is a recent proposal that allows the simultaneous evaluation of several outcomes of interest that can be of any type. These outcomes can be prioritized to reflect one’s opinion regarding their perceived hierarchy of clinical importance. In practice, the method scores the comparison of all possible pairs formed by one patient in the experimental arm and another from the control arm. In case no clear conclusion arises from the comparison for one outcome of interest, the next outcome in the hierarchy is used.

A metric of interest, summarizing the comparisons of all pairs and called ‘Net Treatment Benefit’ (NTB), reflects the difference between the probabilities that a random patient from one group is doing better than a random patient from the opposite group. Such a hierarchical approach empowers patients by formally addressing multiple aspects of their condition and treatment, while explicitly incorporating their opinion.

We will show recent developments of the statistical properties of GPC that now allow the design of clinical trials with the NTB as primary outcome.

  • Samuel Salvaggio, Research Biostatistician, IDDI’s BENEFIT team: Poster Session on: “The Net Treatment Benefit: An Intuitive Statistical Tool Involving Individual Patients in Treatment Decision-Making”
Samuel Salvaggio, Research Biostatistician, IDDI


Personalized medicine requires the incorporation of the patient’s voice in decision-making.

Although a patient-centric approach to decisions has been used in clinical trials, health-technology assessment, and policymaking, typically the perspective mainly reflects decisions and choices made by groups of patients or their advocates. On the other hand, an individual patient-centered decision-making presents additional challenges from the need to take personal context and preference into account. This requires tools that can join a strong methodological basis and a user-friendly interface, with an intuitive presentation.

To tackle these challenges, we propose a new statistical methodology incorporated into an intuitive graphical user interface (GUI) that goes a step further into personalized medicine by allowing individual decisions between two competing therapies that have been tested in a past randomized trial.

Generalized pairwise comparisons is a new statistical method allowing the analysis of the impact of new drugs or health technologies, considering any types and number of outcomes that can be ranked in order of priority. The method provides a clear benefit-risk assessment, because it estimates the net chance of a better outcome with the experimental intervention than with the control treatment. All possible pairwise comparisons between experimental and control treatments are used to compute the Net Treatment Benefit (NTB).

Therefore, this method enables patients to be involved in their own treatment decision-making by allowing them to represent their own hierarchy of outcomes. A data-driven objective metric such as the NTB can also help clinicians to guide patients during decision-making.