Measurement Error of Self-Reported Physical Activity Levels in New York City: Assessment and Correction
Measurement Error of Self-Reported Physical Activity Levels in New York City: Assessment and Correction
复制标题
DOI：
10.1093/aje/kwu470
复制
发表时间：
2015-05-01
影响因子：
5
通讯作者：
Eisenhower, Donna
中科院分区：
文献类型：
--
作者：
Lim, Sungwoo;Wyker, Brett;Eisenhower, Donna
由于很难客观地衡量人口水平的身体活动水平，自我报告的措施已被用作监测工具。然而，很少有人知道他们的有效性在人口稠密的城市地区生活。我们的目的是评估自我报告的身体活动数据的有效性对加速计为基础的测量成年人生活在纽约市，并应用一个实用的工具来调整测量误差在复杂的样本数据使用回归校准方法。我们使用了两个数据组成部分：1）2010-2011年来自3，806名成年人的双帧随机数字拨号电话调查数据，2）来自679名调查参与者的子样本的加速度计数据。自我报告的身体活动水平是使用全球身体活动问卷的一个版本来测量的，而每周中等当量活动分钟数的数据是使用加速度计收集的。两个自我报告的健康指标（肥胖和糖尿病）被纳入作为结果。具有较高加速度计值的参与者更有可能低估实际水平。（加速度计值被视为参考值。）在纠正测量误差后，我们发现结果与身体活动水平之间的关联大大减弱。尽管使用自我报告的数据准确监测密集城市地区的身体活动水平存在困难，但我们的研究结果表明了进行精心设计的验证研究的重要性，因为它允许理解和纠正测量误差。
Because it is difficult to objectively measure population-level physical activity levels, self-reported measures have been used as a surveillance tool. However, little is known about their validity in populations living in dense urban areas. We aimed to assess the validity of self-reported physical activity data against accelerometer-based measurements among adults living in New York City and to apply a practical tool to adjust for measurement error in complex sample data using a regression calibration method. We used 2 components of data: 1) dual-frame random digit dialing telephone survey data from 3,806 adults in 2010-2011 and 2) accelerometer data from a subsample of 679 survey participants. Self-reported physical activity levels were measured using a version of the Global Physical Activity Questionnaire, whereas data on weekly moderate-equivalent minutes of activity were collected using accelerometers. Two self-reported health measures (obesity and diabetes) were included as outcomes. Participants with higher accelerometer values were more likely to underreport the actual levels. (Accelerometer values were considered to be the reference values.) After correcting for measurement errors, we found that associations between outcomes and physical activity levels were substantially deattenuated. Despite difficulties in accurately monitoring physical activity levels in dense urban areas using self-reported data, our findings show the importance of performing a well-designed validation study because it allows for understanding and correcting measurement errors.