Improving Measurement of Physical Activity and Sedentary Behavior
Principal Investigator
Name
Stephanie Eckman
Degrees
Ph.D.
Institution
RTI International
Position Title
Fellow
Email
About this CDAS Project
Study
IDATA
(Learn more about this study)
Project ID
IDATA-50
Initial CDAS Request Approval
Nov 19, 2021
Title
Improving Measurement of Physical Activity and Sedentary Behavior
Summary
As key measures of population health, physical activity (PA) and sedentary behavior (SB) are monitored in many health studies and interventions. Lack of PA contributes to ~10% of premature deaths and is associated with more than $117 billion in annual healthcare costs. PA is an important correlate of health status, and only one in four adults meets federal PA guidelines. Moreover, fewer women, African Americans, Hispanics, and those living in rural areas meet these recommendations. SB is a distinct behavior from PA and is associated with poorer health outcomes including an increased risk for all-cause mortality, cardiovascular disease, and certain cancers.
Many health studies and interventions collect PA and SB data through self-reports; however, self-reports of these behaviors suffer from measurement error because of recall error and social desirability bias. In response to the suboptimal quality of self-reported data, many researchers and expert groups conclude that accelerometry is key to surveillance of physical activity and SB. Accelerometry aims to remove human error and also provides more detail on the components of PA (e.g., frequency, intensity, duration, type) and SB (e.g., bouts, position) than respondents can report. However, accelerometer data are not error free: the devices can lose battery power, and participants may fail to wear them or may change their behavior while wearing them.
Given the central role of PA and SB in public health studies and in the creation of a culture of health and health practices, there is an urgent need to ensure high-quality measurement of these behaviors. When PA and SB are mismeasured, researchers risk misunderstanding their relationship with crucial health outcomes. Interventions to increase PA can be misfocused when PA is poorly measured, with important policy impacts. Furthermore, differences between groups (e.g., ethnicities) in PA and SB can be biased in the presence of differential measurement error. Health researchers need guidance about when to use self-reports of PA and SB and when accelerometer measures are needed. Additionally, there is limited research on how to combine the two types of measures.
Many health studies and interventions collect PA and SB data through self-reports; however, self-reports of these behaviors suffer from measurement error because of recall error and social desirability bias. In response to the suboptimal quality of self-reported data, many researchers and expert groups conclude that accelerometry is key to surveillance of physical activity and SB. Accelerometry aims to remove human error and also provides more detail on the components of PA (e.g., frequency, intensity, duration, type) and SB (e.g., bouts, position) than respondents can report. However, accelerometer data are not error free: the devices can lose battery power, and participants may fail to wear them or may change their behavior while wearing them.
Given the central role of PA and SB in public health studies and in the creation of a culture of health and health practices, there is an urgent need to ensure high-quality measurement of these behaviors. When PA and SB are mismeasured, researchers risk misunderstanding their relationship with crucial health outcomes. Interventions to increase PA can be misfocused when PA is poorly measured, with important policy impacts. Furthermore, differences between groups (e.g., ethnicities) in PA and SB can be biased in the presence of differential measurement error. Health researchers need guidance about when to use self-reports of PA and SB and when accelerometer measures are needed. Additionally, there is limited research on how to combine the two types of measures.
Aims
The project will use multitrait-multimethod models to address three aims:
1. Estimate reliability (variance) and validity (bias) of PA/SB measures from self-reports and accelerometers.
2. Using the results of Aim 1 to corect for measurement error in estimates of PA and SB, estimate the relationship between PA/SB and:
a. important health outcomes such as cardiovascular disease, diabetes, health-related quality of life, all-cause mortality, or self-rated health;
b. sociodemographics such as income, ethnicity, race, age, and gender.
3. Drawing on results of Aims 1 and 2, develop correction factors for self-reported PA/SB for studies without accelerometry data.
Collaborators
none at this time