Affordable Access

Access to the full text

A dynamic stopping rule for phase I clinical trials

Authors
  • Alam, M. Iftakhar
  • Mansur, Mohaimen
Type
Published Article
Journal
Biometrical Letters
Publisher
Sciendo
Publication Date
Jun 01, 2018
Volume
55
Issue
1
Pages
17–30
Identifiers
DOI: 10.2478/bile-2018-0002
Source
De Gruyter
Keywords
License
Green

Abstract

This paper investigates a stopping rule to be utilised in phase I clinical trials. The motivation is to develop a dynamic rule so that a trial stops early if the maximum tolerated dose lies towards the beginning of a dose region. Also, it will employ many patients if the maximum tolerated dose lies towards the end of a dose region. A two-parameter logistic model is assumed for the dose-response data. A trial is stopped early before reaching the maximum number of patients when the width of the Bayesian posterior probability interval of the slope parameter meets a desired value. Instead of setting a pre-specified width to stop at, we determine it based on the parameter estimate obtained after a reasonable number of steps in a trial. Simulation studies of six plausible dose-response scenarios show that the proposed stopping rule is capable of limiting the number of patients to be recruited depending on the underlying scenario. Although the rule is applied to a D-optimum design here, it will be equally applicable to other model-based designs.

Report this publication

Statistics

Seen <100 times