Abstract P069: Temporal Patterns Of Adherence To The Physical Activity Goal In A Mobile Health Weight-loss Trial

Abstract

Background: Wearable trackers such as Fitbit may promote physical activity (PA) levels recommended by the US Physical Activity Guidelines. However, there is a lack of evidence on the associations between long-term wearable tracker use and temporal patterns of PA levels. Therefore, we aimed to identify distinct trajectories of the adherence to PA goals using group-based trajectory modeling in a year-long mobile health weight-loss trial. Method: PA data were collected within the SMARTER weight-loss trial that recruited adults (i.e., age ≥18 years) from 2018 to 2020 from southwest PA, USA. Healthy adults with body mass index between 27-43 kg/m 2 were randomized to either self-monitoring with daily personalized feedback messages (SM+FB, n=251) or self-monitoring only (SM, n=251) and followed for 52 weeks. Adherence to weekly PA goals, the primary outcome of this analysis, was computed as the percentage of fairly active and very active total minutes per week measured by Fitbit Charge 2™ relative to the minimum PA goal of 150 minutes/week recommended for all study participants. We used group-based trajectory modeling to identify distinct temporal patterns of adherence to weekly PA goals. Results: The sample (N = 502) was mostly female (n=399, 79%), white (n=423, 82%) with a mean age of 45.0 ± 14.4 years. Among the models considered, the five-group trajectory model for adherence to weekly PA goals was selected based on the best data fit: low adherence with a linear decline (n=80, 16.0%; b[intercept] ± SE: 73.18 ± 5.03, p<.0001; b[linear] ± SE: -6.37±.28, p<.0001); low initial adherence with nonlinear increase (n=194, 38.7%; b[intercept] ± SE: 93.48 ± 3.63, p<.0001; b[linear] ± SE: .49 ± .31, p=.108;b[quadratic] ± SE: -.03±.01, p<.0001); moderate initial adherence with linear decline (n=145, 28.8%, b[intercept] ± SE: 230.53 ±2.83, p<.0001; b[linear] ± SE: -.40±.09, p<.0001), moderate initial adherence with nonlinear decline (n=39, 7.8%, b[intercept] ± SE: 216.99 ± 18.72, p<.0001; b[linear] ± SE: 0.25 ± 7.46, p=.974; b[quadratic] ± SE: -1.57 ± .93, p=.090; b[cubic] ± SE: .12 ± .05, p=.010; b[quartic] ± SE: -.004 ± .001, p=.010; b[quantic] ± SE: .00003±.00001, p<.0001), and high initial adherence with nonlinear decline (n=44, 8.8%; b[intercept] ± SE: 379.46 ± 9.84, p<.0001; b[linear] ± SE: 2.37 ± 1.59, p=.138; b[quadratic] ± SE: .10 ± .07, p=.171; b[cubic] ± SE: -.003±.001, p=.003). Conclusions: We identified five distinct trajectories for long-term adherence to the PA goal over 52 weeks. More research is needed to determine additional means of supporting PA in individuals with low and declining adherence trajectories.