Ismael Flores Cervantes
Westat
Mr. Ismael Flores Cervantes is a Senior Statistician at Westat.
252 – Replicate Weights and Variance Estimation
Estimating HIV Incidence Using Complex Survey Data
Ismael Flores Cervantes
Westat
J.D. Opsomer
Colorado State University
Rommel Bain
U.S. Centers for Disease Control and Prevention
A. De
U.S. Centers for Disease Control and Prevention
P. Stupp
U.S. Centers for Disease Control and Prevention
Accurate estimation of HIV incidence in at-risk countries is an important public health challenge. The Population-based HIV Impact Assessment (PHIA) Project has been conducting large-scale national population surveys using complex survey designs. HIV incidence estimation for these surveys is performed using a survey-weighted version of the incidence estimator originally proposed in Kassanjee, McWalter, Bärnighausen & Welte (2012). We describe a comprehensive variance estimation approach for this estimator that fully accounts for the variance components due to the survey and those due to the estimation of the biomarker assay parameters. The approach is readily integrated into a large-scale survey context, including that of the PHIA Project surveys. We illustrate the approach on data from three African countries and evaluate the sensitivity of the estimates to the values provided for the biomarker assay parameters and their measures of uncertainty.