Date of Award

January 2013

Document Type

Open Access Thesis

Degree Name

Master of Public Health (MPH)

Department

School of Public Health

First Advisor

Peter Peduzzi

Abstract

Common practice in neoadjuvant therapy clinical trials for breast cancer that use pathological complete response rate (pCR) as an endpoint are conducted within a Human Epidermal Growth Factor Receptor 2 (HER2) indication group for both Estrogen Receptor - positive (ER+) and Estrogen Receptor - negative (ER-) cancers. Given the clinical background and trends of breast cancer therapy trials, this study aims to demonstrate in cases where the observed prevalence and response rates may be so different from the expected values that priori sample size calculation and power analyses were based on, that dangers may arise in producing an insufficiently powered study with unreliable results. Critiques of the widespread practice of underpowered clinical trials are long-standing and such related ethical issues have been substantially debated in biostatistics and medicine. However, an overwhelming prevalence of underpowered studies even recently and studies failing to do a priori sample size and power calculation is still found.

This study uses simple statistical methods to show the effects of not accounting for proportional differences and also detect power differences in pCR rates between two arms in a randomized study. To demonstrate how the overall pCR rate can change for the same effect size in a particular HER2 group study based on changing the proportion of ER+ and ER- patients, pCR rates are calculated over a series of hypothetical studies with varying proportions of cases. The power needed to detect an absolute difference in pCR rates between the two arms could vary greatly depending on the actual trial accrual by ER status.

If a study is designed with specified prevalence rates for ER+ and ER- groups and the observed pCR rates are different than hypothesized, this situation could result in an underpowered study.

Comments

This is an Open Access Thesis.

Share

COinS