Applied Statistics Workshop

Clinical trials are carefully designed research studies conducted to evaluate the safety, effectiveness, and potential side effects of new treatments, drugs, or medical devices. These trials follow a structured protocol, often comparing the new intervention to a placebo or existing standard treatment, to ensure rigorous and unbiased results. They are essential for generating the evidence needed to support regulatory approval and guide medical practice.
Generalized pairwise comparisons (GPC) provide a method to analyze clinical trial data using multiple outcomes of various types (discrete, continuous, or even censored) simultaneously¹. This approach is particularly beneficial when outcomes can be ranked by clinical importance or when specific clinical thresholds are relevant — for example, requiring survival gains to exceed a certain number of months to be deemed meaningful. In randomized clinical trials comparing an experimental treatment to a control, GPC involves pairing each patient from the experimental group with a patient from the control group. These pairs are classified as favorable, unfavorable, or neutral based on the highest-priority outcome. Neutral pairs are subsequently evaluated using the next lower-priority outcome, and this process continues until all outcomes are assessed.
The Net Treatment Benefit (NTB) is calculated as the difference between the proportions of favorable and unfavorable pairs. GPC is especially advantageous as it leverages multiple outcomes^2,3 to enhance statistical power compared to focusing on a single "primary" outcome.⁴ Moreover, its ability to prioritize outcomes flexibly and define clinical relevance thresholds makes GPC a valuable tool for patient-centric analyses.⁴

References
1.    Buyse M. Generalized pairwise comparisons for prioritized outcomes in the two-sample problem. Stat Med 29: 3245-3257, 2010.
2.    Saúde-Conde R, et al. Efficacy and safety of short-course radiotherapy versus total neoadjuvant therapy in older rectal cancer patients: a randomised pragmatic trial (SHAPERS). ESMO Gastrointestinal Oncology, Volume 4, 100067.
3.    Anderson, C, et al. Benefit of Avasopasem Manganese on Severe Oral Mucositis in Head and Neck Cancer in the ROMAN Trial: Unplanned Secondary Analysis. Advances in Radiation Oncology, Volume 0, Issue 0, 101674.
4.    Deltuvaite-Thomas V, De Backer M, Parker S, et al. Generalized pairwise comparisons of prioritized outcomes are a powerful and patient-centric analysis of multi-domain scores. Orphanet J Rare Dis, 2023.

BIO:

Rudradev Sengupta, Senior Trial Design Lead
Rudradev Sengupta is a biostatistician with a Ph.D. in Statistics and over a decade of experience in the pharmaceutical industry, specializing in statistical modeling, advanced analytics, and computing. At One2Treat, he designs patient-centric clinical trials to drive the development of innovative treatments, complementing his dedication to advancing healthcare through scientific publications. Alongside his professional work, he is passionate about teaching and mentoring future biostatisticians at UHasselt.

One2Treat:

One2Treat, founded in July 2023 after four years of incubation within IDDI Group, partners with biopharmaceutical companies to advance clinical trial design, analysis, and market access evaluations. Committed to patient-centricity, it employs modern statistical methods and innovative software to support personalized healthcare solutions.

Teams