Statistical Methods for Analyzing Data of Recurrent Infections after Hematopoieti

描述(由申请人提供)：这个项目的总体目标是开发统计上正确和有效的方法来分析造血细胞移植(HCT)后复发感染之间的间隔时间。感染是HCT后最常见的问题之一，是导致相当大的发病率和死亡率的原因。随着时间的推移，许多患者会反复出现感染性并发症。为了刻画移植后感染并发症的自然历史，识别与感染相关的危险因素，NE需要创新的多元统计方法，能够有效地利用移植中心常规收集的患者和移植相关特征的丰富数据和复发感染事件的时间信息。重复间隔时间数据的现有统计方法 通常假设患者是由于发生与复发事件类型相同的事件而入选的，或者假设包括第一个间隔时间在内的所有间隔时间是相同分布的。将这些方法应用于移植后感染数据，从而忽略事件类型，将不可避免地导致不正确的推断结果，因为从移植到首次感染的时间与首次感染后再感染之间的间隔时间具有不同的生物学意义。或者，人们可能只在第一次感染后分析数据，以使现有的重复间隔时间方法适用，但这引入了选择偏差，因为只有经历过感染的患者才被包括在分析中。其他朴素的方法可以包括对仅第一次感染的数据使用单变量生存分析方法，例如Kaplan-Meier方法和Cox回归模型，或者对直到第二次感染的数据使用双变量生存数据方法，例如Huang-Louis估计器和Lin-Sun-Ying估计器。因此，第一次或第二次感染事件以外的所有后续感染数据将不会在分析中使用，这将导致估计效率低下或功率降低。在这一应用中，我们建议开发有效的统计方法来分析移植后复发感染之间的间隔时间。在特定的目标1中，我们将发展一种非参数估计方法，用于估计感兴趣人群从移植到首次感染的时间和两次感染之间的间隔时间的联合分布。在具体目标2中，我们将建立回归模型，该模型可以结合患者和治疗特点来研究移植患者感染并发症的危险因素。所有建议的方法都将通过广泛的模拟研究和对从2000年1月1日至2010年12月31日期间从明尼苏达大学接受首次HCT的患者收集的现有数据集的分析来评估。将为建议的方法开发用R语言编写的用户友好的程序，并免费提供给公众使用。本研究既具有方法论意义，又具有科学意义。首先，该方法将普遍适用于其他重复间隔时间数据，其中初始事件不同于所有后续事件，这是纵向研究中经常遇到的设计。其次，这项拟议的研究有可能促进对造血细胞移植后感染并发症的自然历史的了解，并有助于识别危险因素和改善患者的移植后护理。

DESCRIPTION (provided by applicant): The overall goal of this project is to develop statistically proper and efficient methods for analyzing the gap times between recurrent infections after hematopoietic cell transplantation (HCT). Infection is one of the most common problems after HCT and accounts for substantial morbidity and mortality. Many patients experience infectious complications repeatedly over time. To characterize the natural history of infectious complications after transplantation and to identify risk factors related to infections, ne needs innovative multivariate statistical methods which can efficiently use the time information of recurrent infectious events and the rich data of patient and transplant related characteristics routinely collected by transplant centers. Existing statistical methods for recurrent gap time data typically assume that patients are enrolled due to the occurrence of an event of the same type as the recurrent event or assume that all gap times, including the first gap, are identically distributed. Applying these methods on the post-transplant infection data, thus ignore event types, will inevitably lead to incorrect inferential results because the time from the transplant t the first infection has a different biological meaning than the gap times between recurrent infections after the first infection. Alternatively, one may only analyze data after the first infetion to make the existing recurrent gap time methods applicable, but this introduces selection bias because only patients who have experienced infections are included in the analysis. Other naive methods may include using the univariate survival analysis methods, e.g., the Kaplan-Meier method and the Cox regression model, on the first infection only data or using the bivariate survival data methods, e.g., the Huang-Louis estimator and the Lin-Sun-Ying estimator, on the data up to the second infections. Hence, all subsequent infection data beyond the first or the second infectious events will not be utilized in the analysis, which will lead to inefficient estimation or a decreased power. In this application, we propose to develop efficient statistical methods for analyzing the gap times between recurrent infections after transplant. In Specific Aim 1, we will develop the nonparametric estimation method for the joint distribution of the time from the transplant to the first infection and the gap times between recurrent infections for the population of interest. In Specific Aim 2, we will develop regression model, which can incorporate patient and treatment characteristics to study the risk factors of infectious complications of the transplant patients. All proposed methods will be evaluated using extensive simulation studies and the analysis of an existing data set collected from patients who received their first HCT from the University of Minnesota between January 1, 2000 and December 31, 2010. User-friendly programs written in R language will be developed for the proposed methods and made freely available for public use. The proposed research holds both methodological significance and scientific significance. First, the methodology will be generally applicable to other recurrent gap time data with the initial event being different from all the subsequent events, which design is frequently encountered in longitudinal studies. Second, the proposed research holds the potential to advance the understanding of the natural history of infectious complications after hematopoietic cell transplantation and to help identify risk factors and improve the post- transplant care of patients.