Enhanced Detection System for Healthcare-Associated Transmission of Infection

项目摘要 尽管最近在减少医疗相关感染（HAI）的发生率方面取得了进展，但 疾病控制和预防部门估计，2015年美国急性护理医院发生了687，000例HAI 并且在给定的一天HAI患病率为30例患者中的1例。估计有72，000名患者死于 在住院期间发生了HAI。此外，医院中的疫情仍然是一个严重的问题， 大多数医院使用过时和无效的方法来检测它们。我们建立了强化者 医疗获得性传播检测系统（EDS-HAT）（R 01 AI 127472），其结合了细菌 全基因组测序（WGS）监测（相对于反应性WGS），以数据检测疫情 电子健康记录（EHR）和机器学习（ML）的挖掘（DM），以确定责任人 传播途径。我们已经证明，EDS-HAT可以检测到两种严重的疫情， 未知和新的传播途径。尽管取得了这一成功，但还需要进行更多的研究， 在EDS-HAT基础上，进一步提高发现和中断医院疫情的能力。例如医院 流感和SARS-CoV-2等呼吸道病毒的爆发有充分的记录，但这一领域 感染预防研究不足。在EDS-HAT中加入呼吸道病毒监测将改善 检测和预防这些昂贵的HAI。此外，可以利用现成的临床微生物学数据， 纳入EDS-HAT算法，以减少对WGS监测的依赖。最后，WGS监控 完全基于核心单核苷酸多态性（SNP）的分析可能错误地将患者聚类。 因此，研究辅助基因组的贡献是必要的，以提高 EDS-HAT的辨别力。在此R 01更新申请中，我们建议利用以下成功案例： EDS-HAT通过开发额外的创新方法来识别和中断医院相关的 传输在目标1中，我们计划使用WGS监测和EHR DM/ML来研究医院传播。 回顾性（aim 1a）和前瞻性（aim 1b）收集的呼吸道病毒阳性样本中的呼吸道病毒 两个大型学术医院（EDS-HAT RV）的样本，一个用于成人，另一个用于儿科。在目标2中，我们 将开发先进的分析方法，以创建一个版本的EDS-HAT，主要依赖于DM/ML的 EHR（EDS-HAT Lite）（目的2a），并提高WGS的区分能力，以正确分类以下患者： 是疾病暴发的一部分（目标2b）。EDS-HAT RV和EDS-HAT Lite将受到临床和预算影响 分析以确定预防的病例数量和避免的医疗费用。这些目标将是 由一个在传染病流行病学、爆发调查、感染 预防，微生物基因组学和基因组流行病学，机器学习和数据挖掘，以及经济 分析和建模。我们提出的研究将导致改善病人的安全性，并可以作为一个模式， 如何在医院里发现和阻止疾病爆发。

Project Summary Despite recent progress in reducing the incidence of healthcare-associated infections (HAIs), the Centers for Disease Control and Prevention estimated that 687,000 HAIs occurred in U.S. acute care hospitals in 2015 and that the HAI prevalence on a given day was one in 30 patients. An estimated 72,000 patients died with HAIs during their hospitalization. In addition, outbreaks in hospitals remain a serious problem but the vast majority of hospitals use antiquated and ineffective methods to detect them. We established the Enhanced Detection System for Healthcare Acquired Transmission (EDS-HAT) (R01AI127472), which combines bacterial whole genome sequencing (WGS) surveillance (as opposed to reactive WGS) to detect outbreaks with data mining (DM) of the electronic health record (EHR) and machine learning (ML) to identify the responsible transmission routes. We have demonstrated that EDS-HAT detects both serious outbreaks that were otherwise unrecognized and novel transmission routes. Despite this success, additional research is needed to improve upon EDS-HAT and further increase capacity to detect and interrupt hospital outbreaks. For example, hospital outbreaks of respiratory viruses such as influenza and SARS-CoV-2 are well documented, but this area of infection prevention is understudied. The addition of respiratory virus surveillance to EDS-HAT would improve detection and prevention of these costly HAIs. In addition, readily-available clinical microbiology data can be incorporated into EDS-HAT algorithms to reduce reliance on WGS surveillance. Finally, WGS surveillance analysis based entirely on core single nucleotide polymorphisms (SNPs) can falsely cluster patients. Therefore, research to investigate the contribution(s) of the accessory genome is necessary to improve discriminatory power of EDS-HAT. In this R01 renewal application, we propose to leverage the success of EDS-HAT by developing additional innovative methods for identification and interruption of hospital-associated transmission. In aim 1, we plan to use WGS surveillance and EHR DM/ML to study hospital transmission of respiratory viruses from retrospective (aim 1a) and prospective collections (aim 1b) of respiratory virus positive specimens at two large academic hospitals (EDS-HAT RV), one for adults and the other pediatric. In aim 2, we will develop advanced analytic methods to create a version of EDS-HAT that relies primarily on DM/ML of the EHR (EDS-HAT Lite) (aim 2a) and improve the discriminatory power of WGS to correctly classify patients who are part of an outbreak (aim 2b). EDS-HAT RV and EDS-HAT Lite will undergo clinical and budget impact analyses to determine the number of cases prevented and healthcare costs averted. These aims will be accomplished by a team with expertise in infectious diseases epidemiology, outbreak investigation, infection prevention, microbial genomics and genomic epidemiology, machine learning and data mining, and economic analysis and modeling. Our proposed research will lead to improved patient safety and can serve as a model for how outbreaks are detected and interrupted in hospitals.