Variability in physical activity patterns as measured by the SenseWear Armband: how many days are needed?

To examine sources of variance in objectively measured physical activity and to determine the number of monitoring days needed to quantify physical activity patterns reliably, 394 Flemish adults (41.1 ± 9.9 years) were monitored during 7 days, using the SenseWear Armband. Differences between weekdays, Saturday and Sunday were examined with repeated measures ANOVA's. Variance components were estimated for subject, weekday and residual error using data from Mondays through Fridays and used to calculate the reliability of 1-5 monitoring weekdays. Saturday was more and Sunday less active than an average weekday. Inter-individual variability was the largest source of variance (54.4-67.9%) for physical activity level (PAL), energy expenditure, inactivity, light, moderate and total physical activity. Intra-individual variability accounted for 31.8-44.8% and weekday for 0.1-1.1% of total variance. Intra-individual variability was the largest source of variance for vigorous activity in both sexes and steps in women. At least, 3 monitoring weekdays were required to achieve a reliability of 0.80 for PAL, energy expenditure, inactivity, light, moderate and total physical activity. All 5 weekdays should be monitored to reach acceptable reliability for steps. Five weekdays resulted in a reliability of 0.58-0.60 for vigorous activity. Both Saturday and Sunday and at least 3 weekdays need to be monitored to obtain reliable measures of habitual physical activity.