Comparing Multiple Approaches to Estimate Physical Activity, Sedentary Behavior, and Sleep in Pregnancy

The purpose of this study was to compare estimates of 24-hour activity using the best practice of a thigh accelerometer (activPAL), wrist actigraphy (Actiwatch), and a sleep diary (PAL + watch + diary) to estimates from simpler procedures, such as the thigh accelerometer and diary (PAL + diary) or thigh monitor alone (PAL only) during pregnancy.

Data collected during the 2^nd trimester from 40 randomly selected participants in the Pregnancy 24/7 cohort study were included. activPAL data were integrated with sleep time determined by wrist actigraphy (PAL + watch + diary) or diary-determined sleep (PAL + diary). In the PAL only analysis, average estimates were exported directly from the PAL software. Repeated measures ANOVA and intraclass correlations coefficients compared moderate-vigorous physical activity (MVPA), light physical activity (LPA), sedentary time, sleep, and wear time across measurement approaches. Pairwise comparisons using a Bonferroni correction explored significant differences identified from the omnibus ANOVA.

The three approaches arrived at consistent durations of physical activity (intraclass correlations coefficients > .95) but not for estimating sedentary behavior and sleep durations (intraclass correlations coefficients: .73-.82). PAL + diary overestimated MVPA by 2.3 min/day (p < .01) compared with PAL + diary + watch. PAL only overestimated sleep (25.3-29.0 min/day, p < .01) while underestimating MVPA (11.7-14.0 min/day, p < .01) compared with the other approaches.

Since the inclusion of the wrist actigraphy provided only slight differences in MVPA estimates, PAL + diary may provide acceptable estimates of 24-hour activity during pregnancy in future research. PAL only may be acceptable when exclusively interested in physical activity.