STATISTICAL METHODS IN AIDS RESEARCH

这笔赠款的总体目标是继续开发和应用 艾滋病研究中的统计方法。有超过15万例艾滋病病例 据疾病控制中心报告，截至1990年10月，艾滋病 流行病是美国最紧迫的公共卫生问题之一 各州。 这项研究的重点是艾滋病的统计建模 流行病。这项提议有三个具体目标。这个 首先是开发统计方法，以获得更好的估计 目前的艾滋病毒流行情况和未来的艾滋病发病率。方法将是 将艾滋病发病数据和潜伏期相结合的调查 获得艾滋病毒流行率的直接估计的分布估计 来自流行病学调查。第二个目标是开发一种 分析系列艾滋病毒流行率调查的统计框架和 确定可以收集到的有关艾滋病毒发病率的哪些信息 从这样的数据感染。该方法将应用于军事领域 在美国筛选申请者数据。第三个目标是仔细 研究可能影响反算的不确定因素 方法以及这些不确定性是如何传播到 对当前和过去艾滋病毒流行率的估计，以及对未来的预测 艾滋病发病率。将考虑的不确定性来源 包括感染率、潜伏期的参数模型1 期间分布、报告艾滋病发病率的延迟和错误，以及 治疗的有效性和有效性。这项工作将需要 详细的灵敏度分析和模型影响的量化 说明有误。我们将描述 多种情况下的反算方法。

The overall objective of this grant is to continue to develop and apply statistical methods in AIDS research. With over 150,000 AIDS cases reported to the Centers for Disease Control as of October 1990, the AIDS epidemic is one of the most urgent public health problems in the United States. The focus of this research is on statistical modeling of the AIDS epidemic. There are three specific objectives of this proposal. The first is to develop statistical methods for obtaining improved estimates of current HIV prevalence .and future AIDS incidence. Methods will be investigated for combining AIDS incidence data and incubation distribution estimates with direct estimates of HIV prevalence obtained from epidemiologic surveys. The second objective is to develop a statistical framework for analyzing serial HIV prevalence surveys and to determine what information can be gleaned about the incidence of HIV infection from such data. The methods will be applied to military applicant screening data in the U.S. The third objective is to carefully study the uncertainties which may effect the back-calculation methodology and how these uncertainties are propagated into the estimates of current and past HIV prevalence, and projections of future AIDS incidence. The sources of uncertainty that will be considered include the parametric mode-1 for the infection rate, the incubation period distribution, reporting delays and errors in AIDS incidence, and the efficacy and availability of treatment. This work will entail detailed sensitivity analyses and quantification of the effects of model misspecification. We will characterize the performance of back-calculation methods In a wide range of situations.