Obesity is a global epidemic with significant health consequences. Sufficient moderate-to-vigorous physical activity (MVPA) is key to preventing obesity, but it represents a small portion of a 24-hour day. Other 24-hour movement behaviors—sleeping, sedentary behavior (SED), and light physical activity (LPA)—also contribute to energy expenditure. Traditional research often examines these behaviors in isolation, overlooking their interdependence. A 24-hour time-use paradigm, examining behaviors as relative components of a day, offers a more comprehensive approach. Compositional data analysis (CoDA) is a suitable method for this, allowing examination of relative changes and the health effects of time reallocations. Previous cross-sectional CoDA studies showed links between higher MVPA relative to other behaviors and lower obesity indicators. However, longitudinal studies examining within-individual changes in 24-h movement behaviors and concurrent changes in obesity indicators are scarce. The transition to retirement provides a natural experiment to study these changes because it often involves altered time use and movement patterns. Prior studies indicate increased sleep duration, decreased physical activity, and increased sedentary time after retirement, sometimes varying by occupational group. Retirement is also linked to increased BMI. This study aimed to investigate how within-individual changes in the composition of 24-h movement behaviors are associated with changes in BMI and waist circumference over one year using the CoDA methodology.

The literature extensively documents the relationship between physical activity, sedentary behavior, sleep, and obesity. Studies highlight the beneficial effects of MVPA and LPA on obesity indicators and the negative association between high SED and obesity risk. Sleep duration's association with obesity is more complex, suggesting a U-shaped relationship with both insufficient and excessive sleep potentially linked to higher obesity risk. However, most studies examine these behaviors independently. The emerging field of time-use epidemiology and the application of CoDA methods offer a powerful approach to analyze the interplay between these behaviors. Previous research using CoDA has shown promising results, but largely lacks longitudinal studies that account for the temporal changes in an individual's movement patterns and the resulting health outcomes. This study addresses this gap by using a longitudinal design in the context of retirement, a life transition known to significantly alter daily movement behaviors.

This study employed a longitudinal design using data from the Finnish Retirement and Aging Study (FIREA). The study population consisted of 213 retiring public sector workers (mean age 63.5 years). Participants wore a thigh-mounted Axivity AX3 accelerometer and completed daily logs for at least four days before and after their retirement transition (approximately one year apart). The accelerometer data, processed using ActiPASS software, provided measures of sleep duration, SED, LPA, and MVPA. BMI and waist circumference were also measured at both time points. Compositional data analysis (CoDA) was applied to analyze the data. The proportion of time spent in each behavior was treated as compositional data, and an isometric logratio (ilr) transformation was used to convert compositional data into real-valued coordinates. Linear regression models examined the associations between changes in 24-h movement behaviors (expressed as ilrs) and changes in BMI and waist circumference, adjusting for baseline measures of movement behaviors, BMI/waist circumference, age, sex, and occupational status (manual vs. non-manual). Isotemporal substitution analysis modeled the effects of reallocating time between different movement behaviors on BMI and waist circumference. Sensitivity analyses were conducted to account for seasonal effects and baseline sleep duration.

The study found a significant association between an increase in MVPA relative to sleep, SED, and LPA and a decrease in BMI (β = -0.60, *p* = 0.04) and waist circumference (β = -2.14, *p* = 0.05) over one year. Conversely, an increase in sleep relative to the other behaviors was associated with an increase in BMI (β = 1.34, *p* = 0.02). Isotemporal substitution analysis revealed that reallocating 60 min from MVPA to SED or sleep increased BMI by 0.8-0.9 kg/m² and waist circumference by 3.0 cm. The effect of decreasing MVPA was larger than the effect of increasing MVPA. Sensitivity analyses adjusting for season and excluding participants with long sleep durations generally supported the primary findings. Increases in LPA relative to other behaviors showed weaker but non-significant associations with reduced BMI and waist circumference. Reallocations between sleep, SED, and LPA resulted in smaller changes in BMI and waist circumference than reallocations involving MVPA.

This study's findings highlight the importance of considering the relative changes in 24-h movement behaviors when examining their impact on obesity, particularly during the transition to retirement. The contrasting effects of increasing MVPA and increasing sleep on BMI and waist circumference underscore the need for tailored recommendations during this life transition. Maintaining MVPA levels, rather than simply aiming for increases, appears crucial given the larger negative effects of MVPA reductions. The observed association between increased sleep and increased BMI may be explained by a decrease in energy expenditure. However, this might be context-dependent; baseline sleep duration levels could influence the association's magnitude. The limited impact of reallocations between sleep, SED, and LPA suggests that these behaviors, having similar low energy expenditure levels, may have less pronounced effects on obesity compared to MVPA.

This longitudinal study using CoDA demonstrated that increasing MVPA relative to other 24-h movement behaviors during retirement transition was associated with decreased BMI and waist circumference, whereas increasing sleep was associated with increased BMI. Maintaining sufficient MVPA levels during retirement is crucial for preventing weight gain and central obesity. Future research should investigate the long-term effects of these behaviors and explore the influence of baseline sleep duration on the associations between sleep changes and obesity. Further research could explore the impact of specific types of physical activity and the role of energy intake.

This study's sample size was relatively small, potentially increasing the risk of type II error. The study primarily included women and non-manual workers, limiting generalizability. The short follow-up period might underestimate the long-term effects of changes in movement behaviors. The study relied on self-reported sleep duration, potentially underestimating actual sleep time. The lack of data on changes in body composition and energy intake represents another limitation. Finally, the study didn't distinguish between different contexts of physical activity.