Basic Principles for Calculating the Required Number of Participants in Clinical Trials. Part 2. Survival Analysis (Review)

• Published in: Regulatory Research and Medicine Evaluation Vol. 15; no. 1; pp. 92 - 104
• Main Authors: Shreder, O. V., Goryachev, D. V., Merkulov, V. A.
• Format: Journal Article
• Language: English; Russian
• Published: Federal State Budgetary Institution ‘Scientific Centre for Expert Evaluation of Medicinal Products’ of the Ministry of Health of the Russian Federation (FSBI ‘SCEEMP’) 01.02.2025
• Subjects: bayesian probability theory; clinical trials; estimation; event-based studies; long-term studies; probability models; sample; sample size; short-term studies; survival
• ISSN: 3034-3062; 3034-3453
• DOI: 10.30895/1991-2919-2025-15-1-92-104

INTRODUCTION. Survival analysis is an important biostatistics method used in clinical trials to confirm the long-term efficacy and safety of medicinal products. The significance of this method lies in the possibility to extrapolate the conclusion regarding the benefits of a medical intervention drawn from a short-term clinical trial to a longer period and adjust dosages and treatment regimens accordingly. However, comprehensive methodological recommendations for planning survival analyses are currently lacking.AIM. This study aimed to systematise the requirements for sample size calculation in event-based study designs.DISCUSSION. This article presents methods for calculating the number of subjects required for survival analysis in event-driven studies with outcome collection and estimation under censoring conditions. The authors discuss Bayesian probabilistic models for estimating survival parameters, such as the time to an event, the risk of an event, and the cumulative survival rate, as key variables for determining the sample size for a study. The article presents a theoretical framework for event risk analysis in survival study designs. The authors describe hypotheses and sta­tistical models for calculating sample sizes and determining survival parameter thresholds in group sequential designs for event risk studies.CONCLUSIONS. The statistical models presented can be used to design studies aimed at estimating the time to an expected event and the cumulative risk during treatment and following medicinal product administration.