Bcrm: Bayesian continual reassessment method designs for phase I dose-finding trials

Michael Sweeting, Adrian Mander, Tony Sabin

Research output: Contribution to journalArticlepeer-review

22 Citations (Scopus)

Abstract

This paper presents the R package bcrm for conducting and assessing Bayesian continual reassessment method (CRM) designs in Phase I dose-escalation trials. CRM designs are a class of adaptive design that select the dose to be given to the next recruited patient based on accumulating toxicity data from patients already recruited into the trial, often using Bayesian methodology. Despite the original CRM design being proposed in 1990, the methodology is still not widely implemented within oncology Phase I trials. The aim of this paper is to demonstrate, through example of the bcrm package, how a variety of possible designs can be easily implemented within the R statistical software, and how properties of the designs can be communicated to trial investigators using simple textual and graphical output obtained from the package. This in turn should facilitate an iterative process to allow a design to be chosen that is suitable to the needs of the investigator. Our bcrm package is the first to offer a large comprehensive choice of CRM designs, priors and escalation procedures, which can be easily compared and contrasted within the package through the assessment of operating characteristics.

Original languageEnglish
Pages (from-to)1-26
Number of pages26
JournalJournal of Statistical Software
Volume54
Issue number13
DOIs
Publication statusPublished - Sep 2013
Externally publishedYes

Keywords

  • Bayesian adaptive designs
  • Continual reassessment method
  • Dose escalation
  • Dose finding
  • Phase I
  • R

Fingerprint

Dive into the research topics of 'Bcrm: Bayesian continual reassessment method designs for phase I dose-finding trials'. Together they form a unique fingerprint.

Cite this