Statistical Issues in Specification of D


Click here to start

Table of Contents

Statistical Issues in Specification of D


The Problem

What is D?

Non-inferiority Trials

Estimation of treatment effect & Confidence Interval

Example 1: Non-inferiority demonstrated

Example 2: Non-inferiority not demonstrated

Goals of non-inferiority trials:

History of D Selection

PPT Slide

Concerns about 1992 PTC approach

Concern of Potential “Bio-Creep”

July 1998 AC Meeting

July 1998 AC Meeting: Proposal

Committee for Proprietary Medicinal Products (CPMP)


Now what?

Road Map

To demonstrate efficacy:

ICH-E10 says


Principles from E10

Possible Choices for D

General approach to D

Estimation of benefit of AC over placebo (D1)

Estimating benefit of AC (D1): Be Conservative

Estimating benefit of AC(D1 ): What is best information?

Estimating benefit of AC(D1): Placebo Control Trial Data

Estimating D1: Bottom Line

D: Step 2

Selection of clinically acceptable loss (D2)

Selection of clinically acceptable loss (D2)

Clinically Acceptable Loss (D2): Consider consequence to patients

Clinically Acceptable Loss (D2): Consider Clinical Trial Realities

Selection of clinically acceptable loss (D2)

General Approach: D for each indication

Consequence to Sample Size


Early Escape: Ethical?

Other Superiority Designs

Selection of D: BIG PICURE

Choice of D impacts patients

Take Home Message