Number of days required to reliably measure weekday and weekend total sleeping time with accelerometer: A secondary data analysis with National Health and Nutritional Survey (NHANES) 2011–2014 data

Abstract

The current standard practice for measuring sleeping time with accelerometer is to ask the participants to wear it for 7 consecutive days and analysing data from participants who have provided at least 4 days of valid data. However, this standard lacks supporting evidence. This study aims to evaluate this standard of practice by examining the reliability of measuring total sleeping time in a representative sample of US adults using accelerometer data from the National Health and Nutritional Survey (NHANES) waves 2011–2012 and 2013–2014. The sample included a total of 14,676 participants, out of which only those who provided data for seven days (n = 9510) were included in the analysis. The results revealed that the intra-class correlation coefficient (ICC) for a single day of measurement was 0.38 for weekdays and 0.27 for weekends. To achieve a reliability of 0.7, measurements for 4 and 7 nights were necessary for weekdays and weekends, respectively. Our simulation study found that the randomly-selected 3-day average of weekday sleeping time strongly correlated with the actual mean (ρ = 0.92), capturing at least 80 % of the variance. However, the randomly-selected 1-day average of weekend sleeping time only captured about 60 % of the variance. In conclusion, we recommend that future accelerometer research adopts a 9-day continuous measurement period, covering four weekend days, to reliably estimate both weekday and weekend sleeping time.