Back to search

Boston University

An approach to conditional power and sample size re-estimation in the presence of within-subject correlated data in adaptive design superiority clinical trials

Abstract

dc:description.abstract

A common approach to adapt the design of a clinical trial based on interim results is sample size re-estimation (SSR). SSR allows an increase in the trial's sample size in order to maintain, at the desired nominal level, the desired power to reject the null hypothesis conditioned on the interim observed treatment effect and its variance (i.e., the conditional power). There are several established approaches to SSR for clinical studies with independent and identically distributed observations; however, no established methods have been developed for trials where there is more than one observation collected per subject where within-subject correlation exists. Without accurately accounting for the within-subject correlation in SSR, a sponsor may incorrectly estimate the trial's conditional power to obtain statistical significance at the final analysis and hence risk overestimating or underestimating the number of patients required to complete the trial as planned. In this dissertation, we propose an extension of Mehta and Pocock's promising zone approach to SSR that reconciles the within-subject correlation in the data for a variety of superiority clinical trials. We consider trials with continuous and binary primary endpoints, and further we explore cases where patients contribute both the same and varying numbers of observations to the analysis of the primary endpoint. Using a simulation study, we show that in each case, our proposed conditional power formula accurately calculates conditional power and our proposed SSR methodology preserves the nominal type I error rate under the null hypothesis and maintains adequate power under the alternative hypothesis. Additionally, we demonstrate the robustness of our methodology to the mis-specification of a variety of distributional assumptions regarding the underlying population from which the data arise.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Biostatistics
Grantor
Boston University
Year dc:date.issued
2022

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Mahoney, Taylor Fitzgerald
Advisor dc:contributor.advisor
  • Massaro, Joseph M.
ORCID dc:identifier.orcid

Subjects

dc:subject × 5

Rights

Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/2144/44807

Chain of custody

source
Harvested from
Boston University
Base URL
open.bu.edu/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Mahoney, Taylor Fitzgerald. An approach to conditional power and sample size re-estimation in the presence of within-subject correlated data in adaptive design superiority clinical trials. doctoral thesis, Boston University, 2022. https://hdl.handle.net/2144/44807