Colliding crises: antimicrobial resistance and ageing

抗菌素耐药性（AMR）正在增加，使感染更难治疗和预防。这一公共卫生危机与人口老龄化相冲突。预计到2050年，世界上超过六分之一的人将超过60岁，与年轻人相比，他们的细菌感染发病率增加了20倍。然而，耐药细菌引起感染的可能性因人的年龄、细菌种类和抗生素而异，目前原因不明。简单分析欧洲感染数据中按年龄对不同抗生素耐药的分离株比例显示，金黄色葡萄球菌（MRSA）对甲氧西林的耐药性在5-18岁年龄组中约为18%，但在65岁及以上的老年人中超过40%。而肠球菌对万古霉素的耐药性在年龄上大致相等，约为3%。我对耐多药结核分枝杆菌感染的研究表明，在全球范围内，15-25岁的人群是感染的高峰。按年龄了解这些趋势很重要，因为这将有助于深入了解抗菌素耐药性选择和传播的潜在机制，从而为改进患者护理和处方方案提供信息（例如，按年龄提供更微妙的抗生素使用建议）。解决这一知识差距需要定量分析，我将在这里首次使用全球数据集系统地绘制按年龄划分的感染抗菌素耐药性模式。我将询问这些模式是否可能是由于临床护理的变化或最近或过去抗生素使用的差异，从而改进抗生素处方和针对抗生素耐药性的干预措施。具体来说，我将对来自全球开放获取数据集的按年龄划分的AMR流行趋势的数据分析与数学模型相结合，以模拟潜在的过程。这种双管齐下的方法将绘制模式图，并利用建模来确定诸如医院传播和耐药细菌携带时间等基本过程的作用。这将需要仔细考虑数据收集过程以及时期和群体效应。要了解抗生素耐药性的这些趋势，还必须考虑不同年龄抗生素使用的变化。然而，全球抗生素使用数据很难找到，而且通常无法按年龄或个体患者水平进行分类。在这个项目中，我将使用一个来自英国的独特数据集，该数据集将抗生素使用与个体患者水平的微生物学信息联系起来，以确定最近（过去一年）与过去抗生素使用对抗生素耐药细菌感染风险的相对重要性。这一点很重要，因为医疗保健使用和抗生素暴露随年龄而变化，因此年龄和抗生素耐药性的任何关联都可能掩盖潜在的抗生素使用趋势。利用这些信息，我将通过研究为模拟潜在过程而开发的数学模型中针对年龄的干预措施的影响，对AMR控制的干预措施进行评估。我还将调整现有的工具，告知经验性抗生素处方，以考虑基于年龄或抗生素使用与抗生素耐药性的关联。经验性处方，即在没有微生物信息的情况下使用抗生素，是全球最常见的抗生素使用方式，因此，在使用中增加细微差别具有减少抗菌素耐药性选择和传播的巨大潜力。通过按年龄剖析AMR的复杂性，本研究将为AMR细菌感染的风险和AMR的起源提供新的见解，以支持基于证据的政策制定。更广泛的影响将是我们对宿主年龄抗菌素耐药性的异质性和趋势的理解的根本性转变，这可能会使患者护理得到长期改善。

Antimicrobial resistance (AMR) is increasing and is making infections harder to treat and prevent. This public health crisis collides with an ageing human population. More than 1 in 6 people in the world in 2050 are predicted to be over 60 years old and they have a 20-fold increase in bacterial infection incidence compared to younger adults. However, the likelihood that infections are caused by resistant bacteria varies by the age of the person, the bacterial species and the antibiotic in a currently unexplained way. A simple analysis of the proportion of isolates resistant to different antibiotics by age in European infection data has shown that resistance to methicillin in Staphylococcus aureus (MRSA) is ~18% in those age 5-18 years old, but over 40% in the elderly (65 years and older). Whereas resistance to vancomycin in Enterococci is roughly equal by age at ~3%. My research on multidrug-resistant Mycobacterium tuberculosis infection suggests that globally there is a peak in infection in those aged 15-25 years old. Understanding these trends by age matters as this would provide an insight into the underlying mechanisms of AMR selection and transmission which could then inform improved patient care and prescribing regimens (e.g., more subtle antibiotic use recommendations by age). Addressing this knowledge gap requires the quantitative analysis I will use here to systematically map, for the first time, AMR patterns in infections by age using global datasets. I will ask whether the patterns could be due to changes in clinical care or to recent or past differences in antibiotic usage, to then improve antibiotic prescribing and interventions against AMR. Specifically, I will pair data analysis of trends in AMR prevalence by age from global open access datasets with a mathematical model to simulate the underlying processes. This two-pronged approach will map the patterns and use modelling to determine the contribution of underlying processes such as transmission in hospitals and duration of carriage of resistant bacteria. It will require careful consideration of data collection processes and both period and cohort effects. To understand these trends in AMR, one must also consider antibiotic usage variation by age. However, global antibiotic usage data is hard to find and is usually not available segregated by age nor at the individual patient level. In this project I will use a unique dataset from England that links antibiotic usage with microbiology information at the individual patient level to determine the relative importance of recent (in the past year) versus past antibiotic usage to risk of infection with an antibiotic resistant bacterium. This is important as healthcare usage and hence antibiotic exposure varies with age, so any association of age and AMR could be masking underlying antibiotic use trends. Using this information, I will then assess interventions for AMR control by investigating the impact of age-targeted interventions in the mathematical model developed to simulate the underlying processes. I will also adapt an existing tool for informing empiric antibiotic prescribing to account for age-based or antibiotic usage associations with AMR. Empiric prescribing, when antibiotics are given in the absence of microbiological information, is the most common use of antibiotics globally and hence increased subtlety in use has enormous potential to reduce AMR selection and spread.By dissecting AMR complexity by age, this research will provide novel insights into the risks of infection with AMR bacteria and the origins of AMR to support evidence-based policy making. The broader impact will be a fundamental shift in our understanding of the heterogeneity and trends in AMR by host age which could enable long-term improvements in patient care.