|
Activity Number:
|
522
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2008 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #302301 |
|
Title:
|
Analysis of Vaccine Adverse Event Count Data with Missing Safety Follow-Up: A Multiple Imputation Method
|
|
Author(s):
|
Xiaoming Li*+ and Jin Xu and Ivan S.F. Chan
|
|
Companies:
|
Merck Research Laboratories and Merck & Co., Inc. and Merck Research Laboratories
|
|
Address:
|
UG1CD-44, North Wales, PA, 19454,
|
|
Keywords:
|
missing data ; safety follow-up ; multiple imputation ; proportion ; incidence rate
|
|
Abstract:
|
Consider a trial in which subjects receive either vaccine or placebo and are followed for serious adverse events (SAE) for 6 months post vaccination. Some subjects may report correlated multiple SAEs and other subjects may not complete the follow-up period, resulting in missing data. One study objective is to compare the relative risk of SAE (vaccine/placebo), and the variables of interest are the proportion of subjects reporting SAE and the incidence of SAE (accounting for follow-up time). A widely used method is the Miettinen and Nurminen method for comparison of two rates. However, this method does not account for missing data and the results may be biased if the data is missing at random. We propose a propensity score-based multiple imputation method as a sensitivity analysis. By simulation, different imputation methods are investigated and compared with the existing method(s).
|
