This repository contains certain supplementary figures, tables and code
Diverse measures of physical activity and variation in health indicators in adults of the United States
Background: Physical activity plays an important role in chronic disease prevention. However, its measurement varies from self-reported questionnaires to objective methods like wearable devices but the impact of these data sources on the bias of downstream analyses is largely uncharacterized.
Methods: Using the participant data from the National Health and Nutrition Examination Surveys (NHANES), we compare self-reported vs. accelerometer physical activity levels using Spearman correlation. We then performed a large systematic association study using multivariate linear regression to relate physical activity with physiological and blood-measured health indicators and quantified the differences in effect sizes between different modes of assessing activity.
Results: We found little to no correlation between self-reported and accelerometer-derived physical activity (max(ρ) = 0.4 (SE)). Our systematic association study revealed that physical activity is associated with multiple health indicators. Additionally, despite the smaller sample size, the use of accelerometer-derived variables resulted in larger absolute effect sizes compared to self-reported variables.
Conclusions: The manner in which physical activity is measured may significantly impact the conclusion of epidemiological studies. In this study, we demonstrate the discordance between self-report and accelerometer-derived physical activity. These discrepancies may lead to imprecise conclusions and even contradictory findings when estimating the association between physical activity and health indicators.