Determining the optimal number of wearing-days given a fixed number of accelerometers in population-level study

Abstract

Background: In research using accelerometers to measure physical activity, the number of accelerometers that can be utilized in a study and the study duration are both constrained. It means that increasing the number of accelerometer wearing days for all subjects leads to a decrease in the total number of participants the study can recruit. We used simulations to find the optimal combination of the number of wearing days and number of participant given a fixed number of accelerometer days. Methods: Two scenarios were studied here, including estimation of population physical activity level and the association between physical activity level and a health outcome. Another similar simulation was conducted by bootstrapping the National Health and Nutrition Examination Survey (NHANES) 2003-2006 accelerometer data (n = 4,069). Results: The simulation results of the first scenario showed that the error was minimized when the number of wearing days was 1 to 2. Simulation results of the second scenario showed that the optimal number of wearing days increased with the total number of accelerometer days and decreased with intra-class correlation (ICC). Conclusion: We developed a tool for researchers to determine the optimal combination of the number of the accelerometer wearing days and the total number of participants and showed that 1 to 2 accelerometer wearing days is optimal for estimation of population physical activity level.