Related Experiment Video
Updated: Mar 9, 2026
Visualization of Intensity Levels to Reduce the Gap Between Self-Reported and Directly Measured Physical Activity
Published on: March 7, 2019
Reproducibility of Accelerometer-Assessed Physical Activity and Sedentary Time
Sarah Kozey Keadle1, Eric J Shiroma2, Masamitsu Kamada3
1Metabolic Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, Maryland; Cancer Prevention Fellowship Program, Division of Cancer Prevention, National Cancer Institute, Bethesda, Maryland.
A 7-day accelerometer measurement reliably captures physical activity and sedentary behavior patterns. This approach is practical for large studies, though adjusting for individual variability can improve chronic disease risk estimates.
Area of Science:
- Epidemiology
- Physical Activity Research
- Biomedical Engineering
Background:
- Accelerometers are increasingly used in large epidemiologic studies.
- Most studies use a single 7-day monitoring period due to cost and logistics.
- The reliability of short-term accelerometer data for estimating long-term behavior patterns is not well understood.
Purpose of the Study:
- To assess the reproducibility of a 7-day accelerometer monitoring period for estimating physical activity and sedentary behavior.
- To determine how well a short-term accelerometer assessment reflects longer-term patterns relevant to disease risk.
Main Methods:
- A subset of 209 women from the Women's Health Study underwent at least two 7-day accelerometer assessments over 2-3 years.
- Data collected included total counts, steps, sedentary time, light-intensity activity, and moderate-to-vigorous physical activity (MVPA).
- Intraclass correlations (ICCs) were calculated for various metrics, adjusting for wear time, age, BMI, and season.
Main Results:
- Intraclass correlations (ICCs) ranged from 0.67 for bouted-MVPA to 0.82 for total daily counts.
- Reproducibility was consistent across different age groups, BMIs, and activity levels.
- Classification accuracy within one quartile was over 90% for all metrics.
Conclusions:
- A 7-day accelerometer protocol provides a reproducible and practical measure of physical activity and sedentary time.
- Variability in ICCs across different metrics suggests a need for careful interpretation.
- Future studies may benefit from methods that adjust for within-person variability in physical activity to refine chronic disease risk estimates.
