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。主要的激励例子是对疫苗有效性的研究。疫苗研究中的两个特别具有挑战性的问题需要评估(I)疫苗接种的间接影响和(Ii)疫苗对感染后终点的影响。评估(I)是一个非标准问题，因为间接影响衡量的是一个人接种疫苗对另一个人健康结果的影响。评估(Ii)具有挑战性，因为受感染的接种者可能无法与受感染的对照相媲美。这项具体建议是调整和发展现代因果推理方法，用于评估(一)和(二)。在出现类似(2)问题的情况下，将进行类似的研究，以防止艾滋病毒从母亲传染给儿童。具体目标1是开发带有干扰的因果推断的统计方法，用于评估直接、间接、总体和总体疫苗效果。特别强调的领域将是发展非参数检验和可信区间，纳入基线协变量，以及对来自观察性研究的数据进行分析。一个鼓舞人心的数据集来自孟加拉国的一项霍乱疫苗试验。具体目标2是开发主要分层因果推断的精确统计方法，用于评估疫苗对感染后终点的效果。本研究将侧重于在最小假设条件下，将主分层的思想应用于小样本环境。将对建议的方法与现有的大样本方法以及传统的意向治疗方法进行比较。这一目标的研究是由疫苗的概念验证临床试验推动的，在这些试验中，预计几乎不会发生什么事件。具体目标3是开发因果推断方法来评估疫苗对传染性的影响。这项研究将结合目标1和目标2的各个方面，因为评估疫苗对传染性的影响的研究通常需要对主要病例的感染进行条件调节(目标2)，而主要病例的疫苗接种状况可能会影响接触密切接触者的感染结果(目标1)。为此目的开发的方法将使用塞内加尔一项研究的数据来评估百日咳疫苗接种对传染性的因果影响。具体目标4是开发具有主要分层和竞争风险的因果推断的统计方法。这项研究的动机是预防艾滋病毒产后母婴传播(MTCT)的研究，在这种情况下，无艾滋病毒死亡和断奶是相互竞争的风险。在这样的研究中，研究人员通常感兴趣的是将以生存为条件的干预策略与特定的时间点进行比较，以便这一目标将利用主要的分层框架。根据这一目标开发的方法将使用马拉维最近一项研究的数据来评估抗逆转录病毒治疗对母婴传播的因果影响。1 公共卫生相关性：这项研究中开发的统计方法将改进对预防传染病干预措施效果的估计。准确和准确地量化干预效果在有关传染病控制的监管决策和公共卫生政策中非常重要。

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. The main motivating examples are studies of vaccine effectiveness. Two particularly challenging problems in vaccine studies entail assessing (i) indirect effects of vaccination and (ii) vaccine effects on post-infection endpoints. Evaluating (i) is a non-standard problem because indirect effects measure the effect of vaccinating one individual on another individual's health outcome. Assessing (ii) is challenging because infected vaccinees may not be comparable to infected controls. This specific proposal is to adapt and develop modern causal inference methodology for use in evaluating (i) and (ii). Similar research will be conducted motivated by studies to prevent transmission of HIV from mother to child where issues similar to (ii) arise. Specific Aim 1 is to develop statistical methods in causal inference with interference for application in evaluating direct, indirect, total, and overall vaccine effects. Areas of particular emphasis will be development of nonparametric tests and confidence intervals, incorporating baseline covariates, and analysis of data from observational studies. A motivating data set is from a trial of cholera vaccines in Bangladesh. Specific Aim 2 is to develop exact statistical methods in causal inference with principal stratification for application in evaluating vaccine effects on post-infection endpoints. This research will focus on applying the ideas of principal stratification in the small sample setting under minimal assumptions. Comparisons will be conducted between the proposed methods and existing large-sample methods as well as traditional intent-to-treat approaches. The research for this aim is motivated by proof-of-concept clinical trials of vaccines where few events are expected. Specific Aim 3 is to develop causal inference methodology to assess vaccine effects on infectiousness. This research will combine aspects of Aims 1 and 2 since studies to assess vaccine effects on infectiousness typically entail conditioning on infection of primary cases (Aim 2) and the vaccination status of the primary case can affect the infection outcome in exposed close contacts (Aim 1). Methods developed in this aims will be used to estimate the causal effect of pertussis vaccination on infectiousness using data from a study in Senegal. Specific Aim 4 is to develop statistical methods for causal inference with principal stratification and competing risks. This research is motivated by studies to prevent postnatal mother-to-child transmission (MTCT) of HIV where HIV-free death and weaning are competing risks. In such studies, investigators are often interested in comparing intervention strategies conditional on survival to a certain time point such that this aim will utilize the principal stratification framework. The methods developed under this aim will be used to estimate the causal effect of antiretroviral therapy on MTCT using data from a recent study in Malawi. 1 PUBLIC HEALTH RELEVANCE: This statistical methods developed in this research will lead to improved estimation of the effects of interventions to prevent infectious diseases. Accurate and precise quantification of intervention effects are important in regulatory decisions and public health policy regarding infectious disease control.