Statistical Methods for Adherence and HIV Outcomes

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Although adherence to antiretroviral medications is increasingly recognized as critical to successful HIV treatment, the development of measurement and statistical methods for adherence lags behind the advancement in development of effective HIV therapies. Studies have shown that adherence to antiretroviral medications is central to suppression of viral replication and to increases in CD4 count, but the exact compliance-response curve and the relationship between adherence and HIV virologic outcomes over time have been inadequately explored. Currently available adherence measures have several major problems. First, there are significant measurement errors or bias associated with each of the measurement techniques. Secondly, most of the measures only measure the percent of prescribed medication taken, ignoring the other aspects of adherence behavior such as dose-interval, which is particularly important for medications with a short half-life such as HIV protease inhibitors. No single adherence measure currently in use integrates the different aspects of adherence behavior. Third, a practical adherence measure feasible for patient care settings is needed. Ease of application and low cost make self-report most appealing for clinical use, but current survey methods fail to gather information with high sensitivity for non-adherence. To date, the relationship between HIV medication adherence and virologic outcomes has been characterized principally using cross sectional analyses with adherence measures that are error-prone and that only partially describe adherence behavior. Using available data from two longitudinal HIV studies that collected rich medication adherence and virologic outcome data, we propose to develop improved methods of collecting and analyzing adherence data, and modeling the relationship between adherence and virologic outcomes. The study will: 1) create measures that can fully and accurately describe the multifaceted character of medication adherence behavior; 2) improve survey methods for obtaining self-report adherence information in the context of clinical care; and 3) using repeated measures analyses, investigate the underlying longitudinal relationship between medication adherence and HIV virologic outcomes of RNA viral load and drug resistance.