|
Activity Number:
|
25
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #303396 |
|
Title:
|
Biological Plausibility-Compliance Average Causal Effect Instead of Per Protocol in the Prevention of Perinatal Sepsis Randomized Control Trial, South Africa
|
|
Author(s):
|
Elizabeth R. Zell*+ and Locadiah Kuwanda and Donald B. Rubin
|
|
Companies:
|
CDC and Respiratory and Meningeal Pathogen Research Unit and Harvard University
|
|
Address:
|
3166 Bolero Drive, Atlanta, GA, 30341,
|
|
Keywords:
|
complier average causal effect ; intent-to-treat ; per protocol ; compliance
|
|
Abstract:
|
Clinical endpoints in randomized trials are commonly evaluated using intent-to-treat (ITT) or PP analyses. ITT is valid for clinical outcomes and estimates the real life effect of the intervention, and is most appropriate when future compliance is like compliance in the trial. PP is often used to estimate biological plausibility but is generally invalid because balance of covariates across treatment groups is generally lost for the subset of participants who meet protocol standards. To evaluate biological plausibility of the intervention, we focus on the subset of true compliers. We illustrate estimation of the CACE for colonization within the PoPs clinical trial designed to evaluate the effectiveness of chlorhexidine vaginal wipes on colonization and neonatal sepsis; the computational method used Markov Chain Monte Carlo with a proper, but relatively diffuse, Bayesian prior.
|
