Use and development of statistical mediation techniques to understand the survival gap between males and females with cystic fibrosis

囊性纤维化（CF）是一种限制生命的遗传性疾病，影响全球超过70，000人。它是由囊性纤维化跨膜调节因子（CFTR）基因突变引起的，该基因为囊性纤维化跨膜传导调节蛋白提供指令，该蛋白充当跨细胞膜运输氯离子和碳酸氢根离子的通道。这种受损的运输导致覆盖细胞的增厚和脱水的粘液，其干扰正常功能并且易于被病原体定殖，从而导致身体的许多器官系统（包括胰腺、肺、肾和消化系统）中的进行性损伤，最终导致寿命缩短目前的治疗集中于减轻CF引起的损害，因为目前没有可以治愈CF的治疗。对于2019年在英国出生的婴儿，男性的中位生存期预计为51.6年，女性为45.7年。近几十年来，即使生存时间延长，男性和女性之间的生存差距也存在于记录的数据中，在其他国家也观察到了这一点。目前还没有明确的解释这种生存差距，它提供了这个项目的动机。似乎性别在生命过程中影响疾病进展的几个方面，这反过来又影响生存。以前的研究表明：女性CF患者在较早的年龄首先获得许多病原体，与男性相比，感染特定病原体的女性死亡率增加，40岁后女性FEV%预测值下降更大（与相同年龄、性别、身高和种族的无CF的人相比，用力呼气量的比例）高于男性，在具有严重CFTR基因型的那些人中，女性具有较高的囊性纤维化相关糖尿病患病率。假设是，这些和其他方面的疾病进展受性别的影响，在组合，负责生存disparity.The项目将使用纵向数据从英国囊性纤维化登记处成立于1995年。这是一个由囊性纤维化信托基金管理的中央数据库。在英国，超过99%的CF患者同意将他们的数据提交给数据库。该数据库是一个大型数据集，不仅包含来自13，000多人的记录和来自年度患者审查的140，000多条记录，而且在每次年度审查时收集的数据超过250个变量的记录的宽度和复杂性方面。该项目将使用截至2018年的数据，其目的之一是通过总结关键变量的性别差异，了解哪些变量显示了性别差异。然而，该项目的主要目的是了解性别对生存的影响如何通过中介变量发挥作用，从而进一步了解导致这种生存差异的因果途径。所考虑的介质将是不同类型的：分类的，连续的和时间到事件，所有的重复测量。最近，调解分析的统计方法有了相当大的发展，其中有多个调解人，而且结果是时间到事件，这个项目提供了一个令人兴奋的机会来使用和发展这些技术。该项目将使我获得以下技能：- 处理和分析来自大型观察性患者登记处的纵向数据-将最先进的统计中介分析和其他因果推理方法应用于实质性研究问题-使用模拟研究开发统计方法和方法评估-与临床顾问和数据专家联络以完善研究问题

Cystic Fibrosis (CF) is a life limiting genetic disorder affecting over 70,000 people worldwide. It is caused by a mutation in the Cystic Fibrosis Transmembrane Regulator (CFTR) gene which supplies the instructions for the cystic fibrosis transmembrane conductance regulator protein which acts as a channel to transport chloride and bicarbonate ions across cell membranes. This impaired transport results in thickened and dehydrated mucus covering the cells that both interferes with normal functioning and is prone to colonisation by pathogens causing progressive damage in many organ systems of the body including the pancreas, lungs, kidneys and digestive system, and ultimately leads to a reduced life-span.Current treatments concentrate on mitigating the damage caused by CF since currently there are no treatments that can cure CF. For babies born in 2019 in the UK, median survival is predicted to be 51.6 years for males and 45.7 years for females. This survival gap between males and females has been present in recorded data over recent decades even as survival times have increased and it has also been observed in other countries. There is as yet no clear explanation for this survival gap and it provides the motivation for this project. It seems likely that sex impacts on several aspects of disease progression through the life course, which in turn affects survival. Previous research has shown that: females with CF first acquire many pathogens at an earlier age than males, females infected with particular pathogens have increased mortality compared to men, after the age of 40 females have greater decline in FEV% predicted (the proportion of Forced Expiratory Volume compared to a person without CF of the same age, sex, height and ethnicity) than males and also that, of those with severe CFTR genotypes, females have higher prevalence of cystic fibrosis related diabetes. The hypothesis is that these and other aspects of disease progression affected by sex are, in combination, responsible for the survival disparity.The project will use longitudinal data from the UK Cystic Fibrosis Registry which was set up in 1995. It is a centralised database managed by the Cystic Fibrosis Trust. Over 99% of people with CF in the UK have consented to their data being submitted to the database. The database is a large data set not only in containing records from over 13,000 people and over 140,000 records from annual patient reviews but also in terms of the width and complexity of the records with data collected on over 250 variables at each annual review. The project will use data up to 2018.One aim of the project is to understand which variables show divergence between the sexes by summarising sex differences in key variables. The main aim of the project, however, is to understand how the effect of sex on survival may act through intermediate variables by using mediation analysis and so further the understanding of the causal pathways leading to this survival disparity. The mediators considered will be of different types: categorical, continuous and time-to-event, and all are repeatedly measured. There has been considerable recent development in statistical methodology for mediation analysis where there are multiple mediators and also where the outcome is time-to-event and this project presents an exciting opportunity to use and develop these techniques.The project will allow me to gain skills in: - Handling and analysing longitudinal data from a large observational patient registry - Applying state of the art statistical mediation analysis and other causal inference methodology to a substantive research question - Developing statistical methodology and assessment of methods using simulation studies - Liaising with clinical advisors and data experts to refine the research questions