Variance Estimation for Directly Standardized Surveillance Rates with Multiply Imputed Race Categories

• Activity Number: 526
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Tracy Pondo*+
• Companies: CDC
• Keywords: surveillance ; overdispersion ; standardized rates ; multiple imputation

Abstract

The Active Bacterial Core surveillance (ABCs) is an active laboratory- and population-based surveillance system for invasive bacterial infections. Race is unknown for about fifteen percent of ABCs cases and the missing race is multiply imputed. After imputation, surveillance rates for three race categories and eight age categories are directly standardized to the age and race distribution of the United States. If the rate of disease is assumed to follow a Poisson distribution, several strategies may be used to calculate interval estimators for directly standardized rates such as the gamma intervals developed by Fay and Feuer, the method introduced by Dobson, Kuulasmaa, Eberle and Scherer, and the normal approximation. We will modify these three interval estimates for standardized rates to incorporate the total variance of the multiply imputed data sets. We propose an additional modification to account for the extra-Poisson variation of the disease rates.