A Measurement Error Approach to Estimating Usual Daily Physical Activity Distribu

描述（由申请人提供）：身体活动评估中的测量误差使得难以回答有关身体活动的流行率以及与各种健康相关结果的关联的重要问题。本申请的目的是开发和评估统计程序，以对常用和公认的身体活动回忆仪器（24小时身体活动回忆24 hPAR）中的测量误差进行建模、量化和调整。为了开发适当的统计模型，我们将进行身体活动测量调查（PAMS），以从居住在爱荷华州农村和城市环境中的1200名成年人（19- 70岁）的代表性样本中获得回忆和客观身体活动数据。参与者将使用BodyMedia Sensewear Pro III（SP3）完成为期两天的身体活动监测，这是一种多通道模式识别设备，可提供PA和能量消耗的准确估计。在每个监测日之后，参与者将完成电话管理的24 hPAR评估。在获得SP3和24 hPAR的重复测量后，将获得自我报告的体力活动倾向数据，以提供用于估计日常体力活动模型和分布的辅助信息。具体目标的连续系列将解决有关体力活动测量误差的独特问题，并导致程序和方法的发展，以促进这些方法在未来研究中的应用。在目标1中，我们将开发一个自我管理的身体活动倾向问卷（PAPQ）和24小时PA回忆（24 hPAR）电话访谈，可以在大规模的调查设置管理。在目标2中，将估计召回和参考测量的测量误差模型，以估计测量的偏倚和随机测量误差结构。所提出的建模的一个独特方面是，我们将利用新的基于倾向的方法来解决人口中许多成年人报告没有体力活动的事实。在目标3中，24 PAR将根据时间匹配的SP3数据进行校准，以便测量结果基本上表现得就像使用参考仪器收集的一样。在目标4中，测量误差模型和校准程序将用于估计亚群中个体的日常体力活动。这项研究的方法是创新的，因为它利用了最先进的监测器，并将导致新的统计技术的发展，以建模和纠正身体活动测量误差。这项拟议的研究意义重大，因为它将直接解决身体活动领域一个复杂而长期存在的测量问题（即获得日常身体活动的准确指标）。这些信息将有助于改善身体活动流行病学研究，并促进更有效的公共卫生监测研究的发展。从这项研究中得到的身体活动测量模型也将促进未来的研究，旨在联合建模能量摄入和能量消耗的误差。该项目由一个强大的研究团队指导，该团队在研究的所有必要方面（身体活动评估，调查设计和管理以及测量误差建模）具有专业知识。公共卫生相关性：拟议的研究将开发和评价统计程序，以模拟、量化和调整常用和公认的体力活动回忆工具（24小时体力活动回忆）中的测量误差。本研究的数据将通过多组分活动监测方案（体力活动测量调查）获得，该方案将从居住在3个种族多样的爱荷华州县的1200名成年人（21-70岁）的代表性样本中收集回忆和客观体力活动数据。数据分析将涉及制定和评价统计程序，根据客观的身体数据校准自我报告措施，以准确估计人口中“通常”的身体活动。

DESCRIPTION (provided by applicant): Measurement error in physical activity assessment has made it difficult to answer important questions about the prevalence of physical activity and associations with various health-related outcomes. The objective in the present application is to develop and evaluate statistical procedures to model, quantify and adjust for measurement error in a commonly used and accepted physical activity recall instrument (24 hour physical activity recall 24hPAR). To develop appropriate statistical models, we will conduct a Physical Activity Measurement Survey (PAMS) to obtain recall and objective physical activity data from a representative sample of 1200 adults (19- 70 yrs) who reside in rural and urban environments in Iowa. Participants will complete two days of physical activity monitoring with the BodyMedia Sensewear Pro III (SP3), a multi-channel pattern recognition device that provides accurate estimates of PA and energy expenditure. After each monitoring day, participants will complete a telephone-administered 24hPAR assessment. After replicate measures of the SP3 and 24hPAR are obtained, self-reported physical activity propensity data will be obtained to provide auxiliary information for estimating models and distributions of usual physical activity. The sequential series of Specific Aims will address unique questions about measurement error in physical activity and lead to the development of procedures and methodologies to facilitate the application of these methods in future research. In Aim 1, we will develop a self- administered physical activity propensity questionnaire (PAPQ) and a 24 hr PA recall (24hPAR) telephone interview that can be administered in a large-scale survey setting. In Aim 2, measurement error models will be estimated for the recall and reference measures in order to estimate the bias and random measurement error structure of measurements. A unique aspect of the proposed modeling is that we will utilize new propensity-based approaches to address the fact that many adults in the population report no physical activity. In Aim 3, the 24PAR will be calibrated against the temporally matched SP3 data so that the measurements essentially behave as if they had been collected using a reference instrument. In Aim 4, the measurement error model and calibration procedures will be used to estimate usual daily physical activity of individuals in subpopulations. The approach in this research is innovative, because it utilizes state of the art monitors and will lead to the development of new statistical techniques to model and correct physical activity measurement error. The proposed research is significant, because it will directly address a complex and long-standing measurement problem in the physical activity field (i.e. obtaining accurate indicators of usual physical activity). This information will help to improve physical activity epidemiology research and facilitate the development of more effective public health surveillance research. The resulting physical activity measurement model from this study will also facilitate future research aimed at jointly modeling error in energy intake and energy expenditure. The project is guided by a strong research team with expertise in all necessary facets of the study (physical activity assessment, survey design and administration, and measurement error modeling). PUBLIC HEALTH RELEVANCE: The proposed study will develop and evaluate statistical procedures to model, quantify and adjust for measurement error in a commonly used and accepted physical activity recall instrument (24 hour physical activity recall). Data for the study will be obtained through a multi-component activity monitoring protocol (Physical Activity Measurement Survey) that will collect recall and objective physical activity data from a representative sample of 1200 adults (21-70 yrs) who reside in 3 ethnically diverse Iowa counties. Data analyses will involve the development and evaluation of statistical procedures that calibrate the self-report measure against objective physical data to obtain accurate estimates of "usual" physical activity in the population.