School-level intraclass correlation for physical activity in sixth grade girls

Area of Science:

• Public Health
• Epidemiology
• Biostatistics

Background:

• Adolescent girls typically experience a decline in moderate to vigorous physical activity (MVPA) as they age.
• Group-randomized trials (GRTs) are a common design for interventions targeting physical activity in this demographic.

Purpose of the Study:

• To report the school-level intraclass correlation (ICC) for MVPA from the Trial for Activity in Adolescent Girls (TAAG) baseline survey.
• To examine how the data collection schedule influences the ICC for MVPA in middle school girls.

Main Methods:

• The TAAG study recruited middle school girls across six sites, with measurements taken using Actigraph accelerometers.
• Data were collected in waves, and moderate to vigorous physical activity (MVPA) was calculated using established criteria.
• Mixed-model regression was employed to estimate the school-level intraclass correlation (ICC).

Main Results:

• School-level ICC for MVPA were higher when data were collected in a single wave versus multiple waves (e.g., 0.057 vs. 0.022).
• ICC estimates also varied between weekdays and weekend days (e.g., 0.024 vs. 0.012).
• Certain data collection schedules demonstrated substantially greater statistical power for detecting intervention effects.

Conclusions:

• The timing and structure of data collection critically influence the ICC for MVPA in adolescent girls.
• This variability can significantly affect the precision of intervention effect estimates and the overall power of a study.
• Researchers must consider the impact of data collection schedules on ICC and MVPA validity when designing future physical activity intervention trials.