Causal Inference in Infectious Disease Prevention Studies

 描述(由申请人提供)：这项研究的总体目标是开发统计方法，以量化预防传染病的干预措施的效果。这项研究的主要激励例子是疫苗研究，尽管所开发的方法将是通用的，并立即在其他环境中应用。疫苗研究中一个特别重要和具有挑战性的问题是评估疫苗接种的间接影响。对于昂贵或在个人接种疫苗时不能完全预防疾病的疫苗，在疫苗引进和使用的政策考虑中，评估间接影响(或群体免疫)是重要的。未能解释群体免疫可能导致关于疫苗对公共卫生益处的错误结论。得出关于群体免疫的推论是非标准的，因为间接影响衡量了一个人接种疫苗对另一个人健康结果的影响。在因果推理的术语中，这被称为“干扰”。也就是说，当一个人的治疗(例如疫苗接种)影响另一个人的结果时，就会出现干扰。在这笔赠款中，将开发创新的统计方法，以便在个人之间可能存在干扰时，对治疗或暴露的影响进行推断。在目标1中，将开发基于随机化(即精确)的统计方法。在目标2中，将开发用于观察性研究的逆概率加权、双重稳健和分层倾向评分治疗效果估计器。目标3将侧重于推断治疗对经过正确审查的事件间隔时间结果的影响。在AIM中，将在不能完全确定因果影响的各种假设下制定处理效果界限和敏感性分析方法。对于目标1-4，将假设个体可以被划分成组，使得不同组中的个体之间不存在干扰； 如果群体在空间上、时间上和/或社会上充分分开，这一假设将是合理的。在目标中，将针对任意形式的干扰开发5种方法 假设种群可以被划分为单独的干扰组。对于所有提议的研究，提议的方法的理论性质将被严格地确立。将进行广泛的模拟研究，以评估所提出的方法在现实环境中的性能。开发的方法将用于分析几项大型传染病预防研究的数据，为不同的 霍乱、流感、肺炎球菌、轮状病毒和伤寒疫苗以及疟疾蚊帐的影响。由此得出的推论将对由于干预而预期的感染数量或避免的疾病病例有直接的解释。所开发的统计方法将适用于许多其他可能存在干扰的环境，包括计量经济学、教育、网络分析、政治学和空间分析。

 DESCRIPTION (provided by applicant): The overall objective of this research is to develop statistical methods for quantifying the effects of interventions to prevent infectious diseases. Th primary motivating examples for this research are studies of vaccines, although the developed methods will be general and have immediate application in other settings. One particularly significant and challenging problem in vaccine studies entails assessing indirect effects of vaccination. For vaccines that are costly or do not afford complete protection from disease when an individual is vaccinated, evaluating the indirect effects (or herd immunity) is important in policy considerations about vaccine introduction and utilization. Failure to account for herd immunity can lead to incorrect conclusions regarding the public health benefit of a vaccine. Drawing inference about herd immunity is non-standard because indirect effects measure the effect of vaccinating one individual on another individual's health outcome. In the nomenclature of causal inference, this is known as "interference." That is, interference is said to be present i the treatment (e.g., vaccination) of one individual affects the outcome of another individual. In this grant innovative statistical methods will be developed for drawing inference about the effects of a treatment or exposure when there is possibly interference between individuals. In Aim 1 randomization-based (i.e., exact) statistical methods will be developed. In Aim 2 inverse probability weighted, doubly robust, and stratified propensity score treatment effect estimators will be developed for observational studies. Aim 3 will focus on inference about treatment effects on time-to-event outcomes subject to right censoring. In Aim 4 treatment effect bounds and sensitivity analysis methods will be developed under various sets of assumptions which do not fully identify the causal effects. For Aims 1 - 4 it will be assumed that individuals can be partitioned into groups such that there is no interference between individuals in different groups; this assumption will be reasonable if the groups are sufficiently separated spatially, temporally, and/or socially. In Aim 5 methods will be developed for arbitrary forms of interference that do not assume the population can be partitioned into separate interference groups. For all of the proposed research, the theoretical properties of the proposed methods will be rigorously established. Extensive simulation studies will be conducted to evaluate the performance of the proposed methods in realistic settings. The developed methods will be used to analyze data from several large infectious disease prevention studies, providing new insights into the different effects of cholera, influenza, pneumococcal, rotavirus, and typhoid vaccines, and malaria bed nets. The resulting inferences will have straightforward interpretations in terms of the expected number of infections or cases of disease averted due to the intervention. The statistical methods developed will be applicable to many other settings where interference may be present, including econometrics, education, network analysis, political science, and spatial analyses.