How many days are needed to estimate accelerometry-assessed physical activity during pregnancy? Methodological analyses based on a cohort study using wrist-worn accelerometer

Background Objective: methods to measure physical activity (PA) can lead to better cross-cultural comparisons, monitoring temporal PA trends, and measuring the effect of interventions. However, when applying this technology in field-work, the accelerometer data processing is prone to methodological issues. One of the most challenging issues relates to standardizing total wear time to provide reliable data across participants. It is generally accepted that at least 4 complete days of accelerometer wear represent a week for adults. It is not known if this same assumption holds true for pregnant women. Aim We assessed the optimal number of days needed to obtain reliable estimates of overall PA and moderate-to-vigorous physical activity (MVPA) during pregnancy using a raw triaxial wrist-worn accelerometer. Methods Cross-sectional analyses were carried out in the antenatal wave of the 2015 Pelotas (Brazil) Birth Cohort Study. Participants wore the wrist ActiGraph wGT3X-BT accelerometer for seven consecutive days. The daily average acceleration, which indicates overall PA, was measured as milli- g (m g ), and time spent in MVPA (minutes/day) was analyzed in 5-minute bouts. ANOVA and Kruskal-Wallis tests were used to compare variability across days of the week. Bland-Altman plots and Spearman-Brown Prophecy Formula were applied to determine the reliability coefficient associated with one to seven days of measurement. Analyses were stratified by sociodemographic factors and nutritional status. Results Among 2,082 pregnant women who wore the accelerometer for seven complete days, overall and MVPA were lower on Sundays compared to other days of the week. Reliability of >=0.80 to evaluate overall PA was reached with at least three monitoring days, whereas six days were needed to estimate reliable measures of MVPA. Conclusions Our findings indicate that the usual approach obtaining one week of accelerometry in adults is also appropriate for pregnant women, particularly to obtain differences on weekend days and reliably estimate MVPA.