Equating NHANES Monitor-Based Physical Activity to Self-Reported Methods to Enhance Ongoing Surveillance Efforts

Abstract

ABSTRACT Purpose Harmonization of assessment methods represents an ongoing challenge in physical activity research. Previous research has demonstrated the utility of calibration approaches to enhance agreement between measures of physical activity. The present study utilizes a calibration methodology to add behavioral context from the Global Physical Activity Questionnaire (GPAQ), an established report-based measure, to enhance interpretations of monitor-based data scored using the novel Monitor Independent Movement Summary (MIMS) methodology. Methods Matching data from the GPAQ and MIMS were obtained from adults (20–80 yr of age) assessed in the 2011–2014 National Health and Nutrition Examination Survey. After developing percentile curves for self-reported activity, a zero-inflated quantile regression model was developed to link MIMS to estimates of moderate to vigorous physical activity (MVPA) from the GPAQ. Results Cross-validation of the model showed that it closely approximated the probability of reporting MVPA across age and activity-level segments, supporting the accuracy of the zero-inflated model component. Validation of the quantile regression component directly corresponded to the 25%, 50%, and 75% values for both men and women, further supporting the model fit. Conclusions This study offers a method of improving activity surveillance by translating accelerometer signals into interpretable behavioral measures using nationally representative data. The model provides accurate estimates of minutes of MVPA at a population level but, because of the bias and error inherent in report-based measures of physical activity, is not suitable for converting or interpreting individual-level data. This study provides an important preliminary step in utilizing information from both device- and report-based methods to triangulate activity related outcomes; however additional measurement error modeling is needed to improve precision.