The effect of a video-supported assessment to increase the accuracy of self-reported physical activity

Abstract

Self-reported physical activity differs from activity levels measured by device. We tested the effect of a video that visualizes the intensity levels of physical activity to increase the agreement between self-reported and accelerometer-based moderate-to-vigorous physical activity (MVPA) within a single-blinded, randomized study. Participants (N = 378, 40-75 years) wore an accelerometer for seven days. Prior to the collection of self-reported data by the IPAQ-SF, participants were randomly assigned (1:1) to a control group (CG) or a video group (VG). The outcome was the absolute difference between self-reported and accelerometer-based time spent in MVPA (Δ MVPAIPAQ−Accelerometry). To examine the agreement, we used Spearman correlation coefficients and Bland-Altman analysis. To test the video effect, we used Wilcoxon signed-rank test, Bayes factor, and simultaneous-quantile regression. In total, 302 participants fulfilled the accelerometer wear time criteria (≥10 hours/day; ≥6 days) and completed self-reports within three days after the wearing period. The median of Δ MVPAIPAQ−Accelerometry was −9.0 min/day (IQR: −32.0 to 66.6) for CG and −11.5 min/day (IQR: −29.9 to 14.3) for VG. Wilcoxon signed-rank test revealed no differences in Δ MVPAIPAQ−Accelerometry between study groups whereas Bayes factor indicated insensitivity of the data. Simultaneous-quantile regression revealed no relationship between video presentation and Δ MVPAIPAQ−Accelerometry in the 25th percentile. In the 50th (b = −12.4 [95% CI = −23.2 to −1.5] and 75th percentile (b = −45.7 [95% CI = −70.5 to −20.9]), Δ MVPAIPAQ−Accelerometry was negatively associated with video presentation. To conclude, video-supported assessment may increase the accuracy of self-reported MVPA among individuals who slightly underestimated and those who overestimated their MVPA.