|
Activity Number:
|
324
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
ENAR
|
| Abstract - #309485 |
|
Title:
|
Effect of Multiple Imputation on the HIV Prevalence Estimate and Analysis of Risk Factors for HIV Infection: Guyana 2004
|
|
Author(s):
|
Maxine Denniston*+ and Nicole Seguy
|
|
Companies:
|
Centers for Disease Control and Prevention and Centers for Disease Control and Prevention
|
|
Address:
|
4757 Banner Elk Drive, Stone Mountain, GA, 30083,
|
|
Keywords:
|
multiple imputation ; survey error ; HIV risk factors ; sample surveys ; HIV prevalence estimation
|
|
Abstract:
|
In response to a report of high levels of HIV infection, a survey of miners in Guyana was fielded using 2-stage cluster sampling to estimate HIV prevalence and risks in this population. In 2006, results were presented from a sensitivity analysis of the effect on the HIV prevalence estimate of uncertainties in survey parameters due to challenges in survey implementation. This paper focuses on the effect of missing data on both the HIV prevalence estimate and risk factors for HIV infection. Multiple imputation was performed on HIV test and risk factor data using Sequential Regression Imputation. Of particular interest was the effect of imputing HIV test results for 35 miners who were interviewed but not tested. Results from this analysis will be compared to sensitivity analysis results. Independent risks for HIV infection will be determined and compared to those from the initial analysis.
|
