CORE--BIOSTATISTICS

点击翻译按钮获取中文摘要

The primary goal of the Biostatistics core is to provide the San Francisco Treatment Research Center (TRC) with support for all of the quantitative and methodological aspects of TRC studies. A secondary goal is to conduct original research of specific statistical issues directly relevant to those same studies. This proposal describes the planned activities of this Core. The core is designed to accomplish the following specific aims: (1) to provide methodological consultation and biostatistical analyses for each component; (2) to provide structure, support, coordination and oversight for collection, computer entry, and management of research data; (3) to insure the security and integrity of all TRC data; and (4) to conduct a series of inter-related studies of existing methods for the analysis of discrete outcomes from longitudinal studies in the presence of missing data. This Core will support the biostatistical and data-related activities of the four proposed components that will study treatments for complex drug-abusing patients in new settings. From a data analysis perspective all component studies include a number of common features: a comparison group, recruiting and enrolling participants sequentially, multiple measures including both continuous and discrete outcomes, participants followed over a time course with repeated measurements, multiple hypotheses, and participant attrition. Each component will also present unique analytic challenges. This Core will perform data entry and data management, and this Core will also maintain the Center computer systems. An integral aspect of the planned Biostatistical Core activities is a program of research focused on an important methodological problem faced by the TRC component projects: choosing the best statistical test for the analysis of data from longitudinal studies using discrete outcomes. Building on current research, computer simulations and real data will be used to study the effects of sample sizes, attrition rates and patterns, and missing data on the behavior of statistical models. Parameter values and data from the individual TRC components will be used in this research so that findings will be directly applicable to the analysis of those projects.