|
Abstract:
|
HIV/AIDS surveillance systems, which collect, analyze, and disseminate data on individuals living with HIVAIDS are used to guide public health action at the federal, state, and local levels. While such data can have noticeable quality problems, such as missing data and measurement errors, they can achieve complete community coverage and produce considerable historical data. Due to privacy concerns surveillance data is often only disseminated in aggregate form. Viral load and CD4 count are two important measures of antiretroviral treatment (ART) responses and HIV disease progression: the proportion of individuals with suppressed viral load (VLS) is used for obtaining reliable estimates of prevention uptake and for assessing the effectiveness of therapy after initiation of ART in a given jurisdiction. We focus on longitudinal measures of VLS and develop statistical methods that integrate surveillance data with sample survey data, to determine which jurisdictions have reliable data, and simultaneously, adjust estimates where bias can be ascertained. We use simulated as well as real data from the enhanced HIV reporting system and the medical monitoring project.
|
