|
Activity Number:
|
45
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 3, 2003 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Social Statistics Section
|
| Abstract - #301259 |
|
Title:
|
Estimating Causal Treatment Effects in a Randomized Encouragement Design with Missing Data
|
|
Author(s):
|
Xiao-Hua Andrew Zhou*+
|
|
Companies:
|
Health Services Research Center-Veterans Affairs Puget Sound Health Care System
|
|
Address:
|
1660 S. Columbian Way, Seattle, WA, 98108-1532,
|
|
Keywords:
|
causal effects ; randomized encouragement design ; maximum likelihood
|
|
Abstract:
|
Because of widely accepted recommendations for influenza vaccination, no controlled randomized trials of the effects of influenza vaccination on pulmonary morbidity in high-risk adults have been published. To around this impasse, McDonald and his colleagues (1978) performed a randomized trial of an intervention that increases the use of influenza vaccine in one group of patients without changing the use of influenza vaccine in another group. This is known as a randomized encouragement design study: the intervention being defined as a computer-generated reminder when a patient with a scheduled appointment was eligible for a flu shot. Estimation of the causal effect of flu shots becomes complicated when the study population is imperfectly compliant and when the outcome is missing for some patient. In this talk, we will discuss an extension of the Rubin Causal Model to handle a binary outcome in a randomized encouragement design study when some patients are missing their outcomes. We will show how to use both the Newton-Raphson algorithm and the EM algorithm to obtain maximum-likelihood estimates of causal parameters.
|
