What Causes Hearing Loss: Advancing the Methods

导致听力损失的原因：改进方法

• 批准号: 10018475
• 负责人: Molin Wang
• 金额: $ 38.31万
• 依托单位: HARVARD SCHOOL OF PUBLIC HEALTH
• 依托单位国家: 美国
• 财政年份: 2019
• 资助国家: 美国
• 起止时间: 2019-09-13 至 2022-08-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/10018475
• 关键词: Accounting; Acquired Deafness; Address; Affect; Applications Grants; Area; Audiology; Auditory Threshold; Clinic; Clinical; Complex; Computer software; Data; Data Analyses; Data Collection; Disease; Ear; Elderly; Environmental Risk Factor; Epidemiologist; Epidemiology; Equipment; Etiology; European; Frequencies; Genetic; Hearing; Hearing Tests; Hospitalization; Individual; Lead; Left; Longitudinal Studies; Measurement; Mental Depression; Methods; Multicenter Studies; National Institute on Deafness and Other Communication Disorders; Nature; Nurses' Health Study; Outcome; Outcome Assessment; Outcome Measure; Outcome Measurement Error; Population; Procedures; Publishing; Quality Control; Quality of life; Research Personnel; Risk Factors; Shapes; Site; Social isolation; Specific qualifier value; Statistical Data Interpretation; Statistical Methods; Structure; Study Section; Testing; Time; Validation; Woman; analytical method; base; burden of illness; cognitive function; cohort; cost; epidemiologic data; epidemiology study; hearing impairment; hearing measurement; modifiable risk; novel; population based; prevent; prospective; research study; software development; structured data; user friendly software; user-friendly; wasting

Project Summary / Abstract Existing statistical methods are insufficient for quality control or analysis of audiometrically-assessed hearing measurements. Statistical analysis of audiometrically-assessed hearing measurements is challenging due to the complex correlation structure of the data, and because the definition of hearing loss is typically based on the pure tone average which is typically the average of measurements at three or four specified frequencies, leading to a potential loss of useful information available in the individual frequency data. In addition, it is now possible to perform audiometric testing outside of the clinic with inexpensive electronic equipment in large- scale epidemiologic hearing studies. In doing so, statistical analysis must account for measurement error in the hearing tests to prevent bias in the estimates of associations and causal effects. We will develop entirely novel methods for quality control of hearing data, and for validly and efficiently assessing the exposure-hearing loss associations and their causal relationships, while accounting for the multiple layers of correlation and multivariate outcomes from multiple frequencies in audiometrically-assessed hearing data. We will develop methods to correct for measurement error-induced bias in the estimated associations and causal effects for studies where hearing outcomes are measured in non-clinical settings. We will apply the hearing data analysis methods to the Conservation of Hearing Study based in the Nurses’ Health Study II. User-friendly publicly available software development will be a central feature accompanying all new methods to be developed. We have formed an interdisciplinary team of expert theoretical and applied statisticians, epidemiologists and audiologists, and we expect to be well equipped to solve the challenging problems that have been identified.

项目摘要/摘要 现有的统计方法不足以控制或分析测听评估听力的质量 测量。由于以下原因，通过测听评估听力测量的统计分析具有挑战性 数据的复杂关联结构，因为听力损失的定义通常基于 纯音平均通常是在三个或四个指定频率上的测量的平均值， 从而导致在各个频率数据中可用的有用信息的潜在损失。此外，它现在是 可以在诊所外使用廉价的电子设备在大型- 规模流行病学听力研究。在这样做时，统计分析必须考虑到 听力测试，以防止在估计关联性和因果关系时出现偏差。我们将开发出全新的 听力数据质量控制和有效评估暴露听力损失的方法 关联及其因果关系，同时考虑到多层关联和 听力测定评估听力数据中来自多个频率的多变量结果。我们将发展 校正估计相关性和因果效应中测量误差引起的偏差的方法 在非临床环境中测量听力结果的研究。我们将应用听证数据分析 以《护士健康研究》为基础的听力保护研究方法II.方便公众使用 可用的软件开发将是伴随所有要开发的新方法的中心特征。我们 已经形成了一支由专家理论和应用统计学家、流行病学家和 我们希望能够很好地解决已经发现的具有挑战性的问题。