Evaluating the tuberculosis epidemic in South Africa using a novel ten-year cohort

项目概要： 结核病（TB）是全球传染病死亡的主要原因。结核病发病率 目前每年减少2%，实现了世界卫生组织的ENDTB目标 要实现到2050年消除结核病的目标，就需要对我们的结核病控制方法进行重大调整。 以前针对其他疾病的消除战略只在空间上成功过一次 确定疾病发生率的变化，然后采取当地相关的干预措施， 切实贯彻消除结核病战略需要纵向结核病队列， 详细的空间信息。南非是我们开展工作的理想场所， 全球结核病发病率最高，国家卫生实验室服务中心（NHLS） 收集的实验室结果数据库。我们建议制定一个十年的国家计划， 结核病队列，能够跟踪患者长达十年，包括空间位置数据 和艾滋病毒状况。我们假设NHLS数据可以用于 在设施一级确定结核病护理级联的关键指标，并评估艾滋病毒的影响 在南非扩大治疗规模和推出结核病流行的新诊断方法。在 目标1a：我们将使用概率性记录链接，在NHLS中创建唯一的患者标识符 所有结核病患者的数据，使我们能够将个人的测试结果联系起来，并跟踪确认的结果 病例的空间和时间，并纳入艾滋病毒检测结果。在目标1b中，我们将确定 结核病护理级联的关键步骤，并使用来自临床图表的数据增强我们的十年队列 评论.在目标1c中，我们将确定在特定护理级联步骤中存在差距的地点， 结核病治疗过程中的患者运动模式，并确定可能更高的位置 基于分子检测变量的疾病负担。在目标2中，我们将量化 艾滋病毒治疗水平和结核病发病率之间的关系。在目标3中，我们将评估 GeneXpert Ultra“跟踪”结果与重复结核病发作之间的关联， 这些新诊断结果的解释。这一贡献意义重大，因为它将 建立一个全国性的结核病患者十年队列，纵向和空间跟踪他们 在治疗期间，使当地公共卫生专业人员能够制定适合当地的 采取干预措施，缩小结核病护理级联中的差距，并了解干预措施的影响 和新的结核病诊断方法。拟议的工作是创新的，因为它将提供第一个 这种高负担环境中的队列，具有这些队列的纵向和空间性质， 这些数据使我们能够前所未有地了解结核病护理和干预措施的影响。这将 使决策者能够针对结核病治疗中的具体差距设计与当地相关的干预措施 和高度移动的人口，防止进一步传播，并最终减少疾病， 结核病死亡率。通过使用常规收集的实验室数据，我们的模型增加了最小的 财务成本，并且可以适应类似的结核病高负担、中等收入环境， 大型实验室数据集。

Project Summary: Tuberculosis (TB) is the leading cause of infectious disease deaths globally. With TB incidence currently decreasing by 2% annually, achieving the World Health Organization’s ENDTB goal of TB elimination by 2050 will require a substantial reworking of our TB control approach. Previous elimination strategies for other diseases have only been successful once spatial variation in disease incidence was identified and then locally relevant interventions implemented. TB elimination strategies require longitudinal TB cohorts that incorporate detailed spatial information. South Africa is an ideal setting for our work with both the second highest TB incidence globally and a central National Health Laboratory Service (NHLS) database of routinely-collected laboratory results. We propose to develop a ten-year national TB cohort with the ability to track patients for up to ten years, including spatial location data and HIV status, using national NHLS data. We hypothesize that NHLS data can be used to identify key metrics in the TB care cascade, at the facility level, and evaluate the impact of HIV treatment scale-up and the roll-out of new diagnostics on the TB epidemic in South Africa. In aim 1a we will use probabilistic record linkage to create unique patient identifiers in the NHLS data for all TB patients, allowing us to link an individual’s test results and track those confirmed cases spatially and temporally, and incorporating HIV test results. In aim 1b, we will identify key steps in the TB care cascade and augment our ten-year cohort with data from clinic chart reviews. In aim 1c, we will identify locations with gaps in specific care cascade steps, common patient movement patterns during TB treatment and identify locations with potential higher burden of disease based on molecular test variables. In aim 2 we will quantify the relationship between HIV treatment levels and TB incidence, at the facility level. In aim 3 we will evaluate the association between GeneXpert Ultra “trace” results and repeat TB episodes to enhance the interpretation of these new diagnostic results. This contribution is significant because it will develop a national ten-year cohort of TB patients tracking them longitudinally and spatially during treatment, enabling local public health professionals to develop locally appropriate interventions to close gaps in the TB care cascade and understand the impact of interventions and new TB diagnostics. The proposed work is innovative because it will provide the first cohort of this kind in a high burden setting, with the longitudinal and spatial nature of these data allowing for unprecedented capture of TB care and the impact of interventions. This will allow policy makers to design locally-relevant interventions targeting specific gaps in TB care and highly mobile populations, preventing further spread and ultimately reducing disease and mortality due to TB. By using routinely collected laboratory data, our model adds minimal financial costs and can be adapted for similar TB high-burden, middle-income settings with large laboratory datasets.