Predicting objective physical activity from self-report surveys: a model validation study using estimated generalized least-squares regression

Abstract

Physical activity measurements derived from self-report surveys are prone to measurement errors. Monitoring devices like accelerometers offer more objective measurements of physical activity, but are impractical for use in large-scale surveys. A model capable of predicting objective measurements of physical activity from self-reports would offer a practical alternative to obtaining measurements directly from monitoring devices. Using data from National Health and Nutrition Examination Survey 2003–2006, we developed and validated models for predicting objective physical activity from self-report variables and other demographic characteristics. The prediction intervals produced by the models were large, suggesting that the ability to predict objective physical activity for individuals from self-reports is limited.

Keywords:

• estimated generalized least squares
• NHANES
• physical activity
• accelerometry
• validation study

Acknowledgements

The authors thank Phil Gleason and the referees for their insightful comments on the manuscript. This work was supported, in part, by the National Institutes of Health under Grant HL091024.

Notes

1. A linear model was used for prediction because we expect a linear relationship between accelerometer and self-reported physical activity; nonlinear models are considered for the purposes of developing the variance model.

2. The purpose of this paper is to see if it is possible to use prediction models for predicting objective physical activity. We focus only on the sample of adult females. A separate paper will focus on samples of adult males and children.

3. The outlier, from the NHANES 2003–2004 sample, was identified as having unrealistic self-reports of physical activity and removed from the sample.

4. As noted earlier, addition self-report variables from the NHANES PAQ measure physical activity related to household activities and average physical activity level on a typical day. We did not include these variables in our final model because the estimated regression coefficients for them were not significant predictors of objective MVPA in any of the models considered during our preliminary analyses.

5. We considered additional models that included more variables but found no difference between these models and the one we develop in this paper based on model comparison tests.

6. Accelerometers, although objective, are still subject to measurement error, due to their inability to capture the full range of activities [20] and the consequence of converting raw accelerometer output into physical activity metrics [18].