Analysis of Multiple Informant Data in Psychiatry

描述（由申请人提供）：在社区或基于服务的心理健康研究中，通常使用多个知情人收集精神病理学和服务使用的评估。例如，在评估儿童的心理健康状况时，父母、教师、儿童和其他人都充当了儿童潜在精神病理学的信息提供者。心理健康治疗的研究和护理质量的评估往往要求多个来源的报告，包括自我报告和行政数据库或医疗图表的查询。精神病学中关于多个线人数据分析的许多方法学挑战仍然没有解决。从1995年发表的工作开始，我们已经开发了一个通用的方法来分析多源数据。虽然这些方法已被证明在各种环境中是有用的，但还需要沿着几条路线进一步发展。该应用程序将开发用于分析多个线人和/或多个评估数据的方法，这些数据是在旨在测量社区和服务样本中的心理健康结果或风险因素的研究中收集的。在具体目标1中，我们将开发更有效的方法来分析用作预测因子的多个线人数据，并开发方法来纳入部分观察到的线人报告。在具体目标2中，我们考虑了在纵向研究中重复测量的多源结局数据的分析方法。在具体目标3中，我们考虑使用多变量潜变量模型来分析非相称的多源结果。我们提出的创新方法将提供一个有吸引力的框架，用于分析多个线人的数据，并允许研究人员系统地研究往往是不同的报告的结构（如精神病理学或服务质量），这是固有的难以衡量。通过将我们的新方法应用于直接适用于最近几项NIMH倡议的现有数据集，我们将向心理健康研究界传播我们的结果。除了描述我们新方法的出版物外，我们还将在现有的多个信息提供者网页上提供宏，文档和教程，并在心理健康和服务会议上提供短期课程。

DESCRIPTION (provided by applicant): Assessments of psychopathology and service use are commonly collected using multiple informants in community or service-based mental health research. For example, in assessments of a child's mental health status, parents, teachers, children and others have served as informants on the underlying psychopathology of the child. Studies of mental health treatment and evaluations of quality of care often solicit multiple source reports, including self-report and queries of administrative databases or medical charts. Many methodological challenges regarding the analysis of multiple informant data in psychiatry remain unsolved. Beginning with work published in 1995, we have developed a general approach to the analysis of multiple source data. While these methods have proven useful in a variety of settings, additional development is needed along several lines. This application will develop methodology for the analysis of multiple informants and/or multiple assessment data collected in studies designed to measure mental health outcomes or risk factors in community and service-based samples. In Specific Aim 1, we will develop more efficient methods for the analysis of multiple informant data used as predictors, as well as develop methods to incorporate partially observed informant reports. In Specific Aim 2, we consider methods for the analysis of multiple source outcome data measured repeatedly in longitudinal studies. In Specific Aim 3, we consider the use of multivariate latent variable models for the analysis of multiple source outcomes that are non-commensurate. The innovative methodology that we propose will provide an attractive framework for the analysis of multiple informant data, and allow researchers to systematically study often disparate reports on a construct (such as psychopathology or quality of services) that is inherently difficult to measure. By applying our new methodology to existing datasets that are directly applicable to several recent NIMH initiatives, we will disseminate our results to the mental health research community. In addition to publications that describe our new methodology, we will provide macros, documentation and tutorials on our existing multiple informant web page, and present short courses at mental health and services conferences.