_Lecture notes_stoppingrules

_Lecture notes_stoppingrules - Data-Dependent Stopping...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Data-Dependent Stopping Rules Phase III studies often are planned with large sample sizes. They make take time to accrue patients and further time to follow for outcome. Interim monitoring may raise questions of whether trial should end early: Early evidence of superiority of treatment in one or more arms. Early evidence of futility (no difference, or new treatment worse). Excess adverse events in one or more arms of study. Poor accrual to study. Focus here is on interim analyses for superiority or futility. Many possible statistical approaches but all share some common features: At study planning stage, investigators define clearly their primary outcome measures, their hypotheses, and what constitutes a meaningful treatment difference. Study design is appropriate for this question: adequate sample size for desired overall power, suitable assignment to treatment, unbiased assessment of outcome. Investigators have agreed on a structure for interim assessment. Typically, the statistical properties of interim stopping rules are fairly similar. In practice, the controversial aspects of decision to stop are not statistical but medical or political. Investigators all recognize that there is a price to be paid for terminating trials early. You may lose power and precision on the primary outcome and on other important study questions, have to compromise on alpha, and/ or have a less persuasive finding. Piantadosi Chapter 14 has a good review. There is an extensive statistical literature. Some references are given below. Four general approaches are mentioned: Fixed sample size approach: keep going until planned sample size is accrued. Frequentist approach: based on hypothesis testing perspective, allow sequential tests. Bayesian approach: requires specifying a prior distribution, then assess posterior dis- tribution and evidence from it....
View Full Document

This note was uploaded on 12/30/2010 for the course BST 252 taught by Professor Tsodikov during the Winter '06 term at UC Davis.

Page1 / 4

_Lecture notes_stoppingrules - Data-Dependent Stopping...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online