The rising prevalence of obesity, fueled by sedentary lifestyles and technological advancements, necessitates a deeper understanding of the relationship between movement behaviors and BMI. While previous studies have focused on sedentary behavior and moderate-to-vigorous physical activity (MVPA), this research expands the investigation to include sleep and light physical activity (LPA), acknowledging the time-constrained nature of these behaviors. Self-reported data, commonly used in large-scale epidemiological studies, are susceptible to measurement errors. Therefore, this study leverages Compositional Data Analysis (CoDA) to analyze the concurrent relationships between sleep, SED, LPA, MVPA, and BMI using both a self-reported method (24-hour Physical Activity Recall, 24PAR) and an objective monitoring method (SenseWear Armband, SWA). The study aims to determine the associations between these movement behaviors and BMI using CoDA and to compare the associations obtained from these two different measurement methods in a representative sample of adults.

Existing research establishes links between lifestyle movement behaviors and obesity, but often overlooks the time-constrained nature of these behaviors. Studies show that excessive sedentary time negatively impacts health outcomes, regardless of MVPA levels, and that obese individuals tend to be more sedentary and less active. However, research on sedentary behavior is still evolving, facing challenges in measurement and definition. While some studies have investigated the relationship between sedentary behavior and obesity using both self-report and monitor-based measures, the findings remain inconsistent, potentially due to limitations in data processing methods and the use of long-term recall measures. CoDA offers an advantage by accounting for the compositional properties of time-based behaviors, allowing the impact of time allocation on weight status to be evaluated.

This ancillary study utilizes data from the Physical Activity Measurement Survey (PAMS), a cross-sectional study involving a representative sample of 1247 adults (aged 20–75) from four Iowa counties. Participants underwent two replicate trials, each consisting of 24-hour activity monitoring using the SenseWear Armband (SWA) followed by a telephone-administered 24-hour Physical Activity Recall (24PAR) the next day. The SWA, a multi-sensor device, provides various activity parameters, while the 24PAR assesses activity time, energy expenditure, and activity context through a structured interview. Data were processed to classify each minute into SED (≤1.5 MET), LPA (1.5 < MET < 3.0), and MVPA (≥3.0 MET) categories. Sleep time was subtracted from total sedentary time to obtain the final SED value for both methods. CoDA was employed, converting minutes per day into percentages of wear time for each behavior, and log-ratio data transformations (Isometric Log Ratios, ILRs) were performed to analyze the data in real space. Multiple linear regression models were used to assess the relationship between ILRs and BMI, adjusting for several covariates (gender, age, ethnicity, income, employment, education, marital status, smoking status, and measurement day of the week). Four permutations of ILRs were calculated to obtain parameter estimates for each of the four compositional parts (SLEEP, SED, LPA, MVPA). Compositional isotemporal substitution was used to interpret parameter estimates, determining how reallocating time between behaviors impacts BMI. Statistical significance was set at p < 0.05, and analyses were performed using STATA v15.0.

The study found significant associations between time spent in sleep, sedentary behavior (SED), and moderate-to-vigorous physical activity (MVPA) and BMI, using both the 24PAR and SWA methods. However, the SWA models explained a substantially larger proportion of variance in BMI (R² = 0.28) compared to the 24PAR models (R² = 0.07). Specifically, using 24PAR, time spent in sleep (γ = −3.58, p = 0.011), SED (γ = 3.70, p = 0.002), and MVPA (γ = −0.53, p = 0.018) were associated with BMI. Using SWA, time spent in sleep (γ = −5.10, p < 0.001), SED (γ = 8.93, p < 0.001), LPA (γ = −3.12, p < 0.001), and MVPA (γ = −1.43, p < 0.001) were associated with BMI. Compositional isotemporal substitution models revealed that replacing sedentary time with MVPA, LPA (in SWA models only), or sleep was associated with reductions in BMI, with greater effects observed using the SWA method. Significant differences were observed between obese and non-obese participants in terms of income, education, and smoking status.

The findings highlight the importance of considering the relative time spent in various movement behaviors when examining their association with BMI. The stronger associations observed using the objective SWA method compared to the self-reported 24PAR method emphasize the potential for measurement error in self-report instruments. The results support the need for interventions focused on reducing sedentary time and increasing both moderate-to-vigorous and light-intensity physical activity. The significant differences observed in income, education, and smoking status between obese and non-obese participants suggest the need for targeted interventions considering these socio-economic factors. The study's findings contribute to a more nuanced understanding of the complex relationships between movement behaviors and obesity, informing the development of effective interventions.

This study demonstrates the robust association between relative time spent in various movement behaviors and BMI, with the monitor-based SWA providing stronger associations than the self-reported 24PAR. Replacing sedentary time with other behaviors like MVPA, LPA, or sleep is associated with lower BMI. Future research should focus on longitudinal studies to establish causal relationships and explore the effectiveness of interventions targeting specific behavior substitutions, accounting for socioeconomic factors.

The cross-sectional design limits the ability to establish causal relationships between movement behaviors and BMI. The study is limited to a sample from Iowa, and the generalizability to other populations may be limited. The reliance on a specific activity monitor (SWA) and self-report instrument (24PAR) may affect generalizability to other measurement tools. Finally, the study focuses on only four primary behaviors, while other behaviors could potentially be related to BMI.