Search

More time in moderate-to-vigorous physical activity (MVPA) is associated with lower cardiovascular disease (CVD) risk. However, most of the 24-hour day is spent in other behaviours, including light physical activity, sedentary behaviour, and sleep. Device-based movement measurement in large prospective cohorts now enables detailed investigation of 24-hour movement behaviours. However, traditional approaches to classifying behaviours in device data, based on ‘cut-points’, can be crude and sometimes inaccurate. Furthermore, approaches to associate 24-hour movement behaviours with disease risk do not always account for how behaviours make up the 24-hour day. While biases are known to affect analyses, they are rarely assessed quantitatively. This work therefore aimed to investigate effects of device-measured 24-hour movement behaviours on risk of incident CVD, addressing challenges around behaviour classification, analysis of 24-hour movement behaviours, and interpretation. This work developed machine-learning-based classification methods to classify free-living wrist-worn accelerometer data as sleep, sedentary behaviour, light physical activity behaviours, or moderate-to-vigorous physical activity behaviours. The classifier used a Random Forest with a Hidden Markov model for time smoothing. It was developed using free-living data from a study of 152 participants who had a camera-and diary-based ground-truth measured alongside accelerometry (the CAPTURE-24 study). In Leave-One-Participant-Out analysis, this method classified free-living movement behaviours with a mean accuracy of 88% (87%, 89%) and a mean Cohen’s kappa of 0.80 (0.79, 0.82). UK Biobank (UKB) is a population-based prospective cohort of middle-to-older aged adults, in which∼100,000 participants wore a wrist-worn Axivity AX3 accelerometer for a 7-day period between 2013 and 2015. The machine-learning model was used to classify movement behaviours of UKB participants, and classification patterns by time of day and by participant characteristics showed good face validity. UK Biobank participants’ 24-hour movement behaviours were associated with risk of incident cardiovascular disease (ICD-10 codes I20-25, I60-69) using a multivariable-adjusted Cox regression model. To account for the compositional nature of 24-hour movement behaviour data, it was modelled using a Compositional Data Analysis approach. Over a median of 6.2 years’ follow-up, there were 4,105 incident CVD events among 87,498 participants. Reallocating time from other behaviours to MVPA was associated with a lower risk of incident CVD. Reallocating time from sedentary behaviour to other behaviours was also associated with lower risk. For example, compared to the average movement behaviour composition in this cohort, reallocating 20 minutes/day to MVPA from all other behaviours proportionally was associated with a Hazard Ratio (HR) of 0.91 (0.90, 0.93), while reallocating 1 hour/day to sedentary behaviour from all other behaviours proportionally was associated with a HR of 1.05 (1.03, 1.07). A range of approaches were used to quantitatively assess the impact of information bias, selection bias, reverse causality, and residual/unmeasured confounding on results. Adjusting for within-person variation in measurements modestly de-attenuated some results and little affected others. Although there were clear selection effects from UK Biobank into the accelerometry sub-study, this did not appear to substantially affect analytic results. Findings on reverse causality were mixed, with little difference by time period but different findings on sedentary behaviour by (cardiovascular) health status. There was some evidence residual/unmeasured confounding affected MVPA results, but it did not entirely explain findings. This work contributes to emerging evidence on device-measured 24-hour movement behaviours and risk of incident cardiovascular disease. It both supports existing guidelines encouraging adults to engage in regular moderate-to-vigorous physical activity and limit sedentary time, replacing it with physical activity of any intensity, and supports the development of future guidelines and interventions targeting behaviours throughout the 24-hour day. This work also supports future research, particularly by developing and making available a machine-learning-based classifier with good performance, and by using emerging methods for associating movement behaviours with incident disease and for assessing bias.

Reallocations of time between daily activities such as sleep, sedentary behavior and physical activity are differentially associated with markers of physical, mental and social health. An individual’s most desirable allocation of time may differ depending on which outcomes they value most, with these outcomes potentially competing with each other for reallocations. We aimed to develop an interactive app that translates how self-selected time reallocations are associated with multiple health measures. We used data from the Australian Child Health CheckPoint study (n = 1685, 48% female, 11–12 y), with time spent in daily activities derived from a validated 24-h recall instrument, %body fat from bioelectric impedance, psychosocial health from the Pediatric Quality of Life Inventory and academic performance (writing) from national standardized tests. We created a user-interface to the compositional isotemporal substitution model with interactive sliders that can be manipulated to self-select time reallocations between activities. The time-use composition was significantly associated with body fat percentage (F = 2.66, P < .001), psychosocial health (F = 4.02, P < .001), and academic performance (F = 2.76, P < .001). Dragging the sliders on the app shows how self-selected time reallocations are associated with the health measures. For example, reallocating 60 minutes from screen time to physical activity was associated with -0.8 [95% CI -1.0 to -0.5] %body fat, +1.9 [1.4 to 2.5] psychosocial score and +4.5 [1.8 to 7.2] academic performance. Our app allows the health associations of time reallocations to be compared against each other. Interactive interfaces provide flexibility in selecting which time reallocations to investigate, and may transform how research findings are disseminated.