Understanding the nature of measurement error when estimating energy expenditure and physical activity via physical activity recall

Authors: PAUL, DAVID - University Of Idaho; MCGRATH, RYAN - University Of Idaho; VELLA, CHANTAL - University Of Idaho; Kramer, Matthew; Baer, David; Moshfegh, Alanna

Submitted to: Journal of Physical Activity and Health
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/2/2018
Publication Date: 3/26/2018
Citation: Paul, D.R., McGrath, R.P., Vella, C.A., Kramer, M.H., Baer, D.J., Moshfegh, A.J. 2018. Understanding the nature of measurement error when estimating energy expenditure and physical activity via physical activity recall. Journal of Physical Activity and Health. https://doi.org/10.1123/jpah.2017-0089. PMID: 29580135.
DOI: https://doi.org/10.1123/jpah.2017-0089

Interpretive Summary: Engaging in physical activity has been shown to be related to a number of health benefits including lowering the risk of weight gain, metabolic syndrome, diabetes, and heart disease in adults. However, despite the importance of investigating the health aspects of physical activity, scientists continue to struggle with the complexities associated with accurately quantifying it. The nation’s premier health survey, the National Health and Nutrition Examination Survey (NHANES), includes a Physical Activity Questionnaire (PAQ) that is administered to participants. The results from the NHANES PAQ administered to adults participating in the USDA AMPM Validation Study were used to compare energy expenditure and physical activity from the PAQ to those from doubly-labeled water (DLW) and accelerometers. Results showed that the use of estimates from the NHANES PAQ data provided comparable results for activity energy expenditure (AEE) when compared to DLW estimates, but significantly overestimated moderate-to-vigorous physical activity (MVPA) when compared to accelerometer data. Thus, use of estimates from the NHANES PAQ may result in underestimates of the true relationship between AEE and MVPA, and disease.

Technical Abstract: This study investigated differences in the estimates of Activity Energy Expenditure (AEE) and Moderate-to-Vigorous Physical Activity (MVPA) from the National Health and Nutrition Examination Survey (NHANES) Physical Activity Questionnaire (PAQ), when compared to estimates from doubly-labeled water and accelerometers. This study also investigated the bias and variance of the errors in the estimates of AEE and MVPA when utilized to predict health markers and categorize individuals. The PAQ was administered to 524 adults to estimate AEE (AEE-PAQ) and MVPA (MVPA-PAQ), while simultaneously measuring AEE via doubly labeled water (AEE-DLW) and MVPA via accelerometers (MVPA-A). Body composition was measured via dual-energy X-ray absorptiometry and resting energy expenditure was measured via indirect calorimetry. Blood samples were collected and assayed for several health markers. AEE-PAQ was not significantly different from AEE-DLW (P=0.13) despite a mean absolute difference of 2.1 MJ/d. However, MVPA-PAQ was significantly higher than MVPA-A (P<0.0001), with a mean absolute difference of 22.0 min/day. Regression analyses between AEE-PAQ/AEE-DLW, and MVPA-PAQ/MVPA-A yielded significant positive relationships, with residual variances of 2.5 and 333.3, respectively. Evidence for attenuation when using the PAQ to predict health markers was observed in 8/12 cases when compared to doubly-labeled water and accelerometers. When compared to accelerometers, the PAQ correctly identified 77.1% and 37.7% of participants that met or did not meet American College of Sports Medicine (ACSM) recommendations for physical activity, respectively. Although the group-level estimate error in AEE was small, the NHANES PAQ significantly overestimated MVPA. Poor within-subject estimates of AEE and MVPA lead to attenuated predictions of various health markers and misclassifications of individuals that met or did not meet ACSM recommendations for physical activity.