Physical activity in adolescents – Accelerometer data reduction criteria: Number of valid days, minimum daily wear time, and non-wear time

Introduction: Accelerometry is increasingly being recognized as an accurate and reliable method to assess free-living physical activity (PA).
However, reporting of accelerometer data reduction and methods remains inconsistent. In this study we investigated the impact of different data
reduction criteria on participant inclusion and PA data outcomes.
Methods: We used data from the Danish SPACE for physical activity study (n=1,348,11–13y). Adolescents wore the Actigraph GT3X for seven
consecutive days. Accelerometer data were analyzed using a range of values for three key data reduction issues: number of valid days (1, 2, 3, 4,
5, 6 and 7days), daily wear time (6, 8, 9, 10 and 12 h/day) and non-wear time (10, 20, 30, 60 and 90 min of consecutive zeroes). The open source
software Propero Actigraph Data Analyzer was used to compare the effects of the selected criteria on participant inclusion and PA outcomes
(mean cpm). The following parameters in the data reduction analyses were fixed: 30sec epoch, 24h duration, first registration accepted after 4h,
maximum value 20,000cpm, and two activity epochs permitted in blocks of non-wear.
Results: Accelerometer data were obtained from a total of 1,296 adolescents. Descriptive analyses showed that increasing minimum daily wear
time and number of valid days resulted in a lower percentage of participants included for analysis. On average, 98.3% of participants had at least
1 valid day, 90.6% had 4 days, and 51.3% had 7 days. Lengthening non-wear duration resulted in a higher percentage of participants included.
In general we found the most substantial differences in compliance when looking at 10–12h (daily wear time), and 5–7days (number of valid days).
Only 4.2% of participants had 7 valid days of 12h wear time, whereas 98.8% of participants had at least 1 valid day of 6h wear time using a 10min
non-wear criterion. The corresponding numbers using a 90min non-wear criterion were 20.6% and 99.4%. Lengthening the non-wear period
decreases PA level (mean cpm) substantially, e.g. average PA was 641 cpm (5 days of 10h) using the 10min non-wear criterion compared to
570 cpm using 90min non-wear. No systematic differences in PA outcomes were found when comparing the range of days and hours.
Discussion: We used a systematic approach to illustrate that even small inconsistencies in accelerometer data reduction can have substantial
impact on compliance and PA outcomes. Optimal data processing techniques will depend significantly on the research question to be answered.
Support: TrygFonden supported the project.