The “Sweet-Spot Hypothesis” postulates that the best balance between 24-h physical behaviors for better health differs between adults in sedentary and physically active occupations. Specifically, testing this hypothesis is about falsifying the null hypothesis (H0) that (1) there is no difference in the “Sweet-Spot” of 24-h physical behaviors between adults in occupations with different physical behaviors, and (2) the advice “sit less–move more” brings all adults from different occupations toward their “Sweet-Spot” of 24-h physical behaviors for better health.
We suggest the following steps to test the “Sweet-Spot Hypothesis.” Firstly, socioeconomic position is closely linked to both 24-h physical behaviors and health, and thus, it is important to sufficiently account for socioeconomic confounding. One way to do so is through study design, by ensuring study populations with considerable variance in physical behaviors at work but with a homogenous socioeconomic position. In any case, we encourage researchers to collect information on socioeconomic position (e.g., education, income and occupation).
Secondly, considering that self-reported measures of daily physical behaviors are susceptible to misclassification bias [28], we recommend using device-based measurements of 24-h physical behaviors. Moreover, we recommend supplementing device-based measures with self-reports or other means to ascertain the context of the physical behaviors (e.g., work, recreational, transport and domestic), as well as other conditions of the work behaviors including level of control over work tasks and amount of lifting/loading. Other important aspects of 24-h physical behaviors to consider may include day-to-day variability in behaviors (is there a consistent routine?); the timing of activities (are the most physically demanding tasks early in the morning or later in the afternoon); bout distribution (e.g., is sitting time interspersed with active breaks? Is sleep only nocturnal, or is there napping during the day); and the quality of behaviors (e.g., is MVPA from lifting heavy objects or running? Is sleep regularly disturbed?). Additionally, information on potential confounders, such as other lifestyle behaviors, physical and mental health status should be collected.
Thirdly, it is important to acknowledge that daily physical behaviors are parts of a finite whole where increasing time spent in one behavior necessarily means less time for other behaviors. Therefore, data on time spent in physical behaviors are compositional in nature, conveying relative—rather than absolute—information [29]. Accordingly, assessment, reporting of results and interpretation of the association between physical behaviors and health should be in relative terms, considering the co-dependency of daily physical behaviors [30, 31]. We recommend the use of analytical methods that address the nature of compositional data, such as compositional data analysis (CoDA), for analyzing relationships between times spent in physical behaviors and health [32, 33].
An Example of Testing the “Sweet-Spot Hypothesis”
Figure 1 is a ternary diagram illustrating the cross-sectional association between 24-h compositional physical behaviors and self-rated health of adults in predominantly desk-based occupations (“white-collar,” Fig. 1A), manufacturing occupations (Fig. 1B) and cleaning occupations (Fig. 1C) from the DPhacto cohort [34]. We used 24-h thigh-worn accelerometry and the Acti4 software [35] to estimate the daily time spent sedentary (lying and sitting), active (standing, walking, running, cycling and stair climbing) and “in bed” (as proxy for sleep based on participants’ diary information). Following a CoDA approach, we tested the interaction between physical behaviors (transformed to isometric log-ratios) and occupation against self-rated health, using a second-order polynomial model [36]. The resulting interaction tended to be significant (P = 0.06), indicating differences between occupations in the association between 24-h compositional physical behaviors and self-rated health. To understand these differences, using the model estimates, we predicted the compositions of 24-h physical behaviors associated with the best 5% (defined as the “Sweet-Spot,” illustrated by the dark green areas in Fig. 1), 5–10%, and 10–15% of self-rated health within each occupation, adjusted for age, sex, body mass index and smoking.
Figure 1 visualizes the sweet-spot (green-colored areas) for adults in different occupations and shows how this “Sweet-Spot” compares to each occupation’s average daily 24-h physical behavior composition (black dot). The black arrows show the direction of physical behavior change required to bring each occupation closer toward their “Sweet-Spot.” The red arrows show the direction of physical behavior change recommended by the simple ‘move more-sit less’ recommendation.
Figure 1A shows that for adults in white-collar (administrative, mainly sedentary) occupations, the 24-h physical behavior distribution associated with the best 5% of self-rated health comprised about 30% of the day spent on sedentary behavior, 45% spent actively, and 25% spent on sleep. Therefore, the “Sweet-Spot” of 24-h physical behaviors associated with the best self-rated health appear to be achieved by following the “sit less–move more” advice, indicated by the overlapping red and black arrows from the mean composition of physical behaviors for the occupational group.
For adults working in manufacturing, Fig. 1B shows that the 24-h physical behavior distribution associated with the best 5% of self-rated health comprised about 35% spent sedentary, 35% spent actively, and 30% spent on sleep. Therefore, for adults in this occupational group, the “Sweet-Spot” of 24-h physical behaviors associated with the best self-rated health was not achieved by following the “sit less–move more” advice (indicated by the red arrow), but rather, by increasing sedentary and active time while decreasing sleep time (indicated by the black arrow).
For the adults who work in cleaning, Fig. 1C illustrates that the 24-h physical behavior distribution associated with the best 5% of self-rated health comprised about 50% spent sedentary, 15% spent actively, and 35% on sleep. For the adults in this occupation, the “Sweet-Spot” of 24-h physical behaviors was not achieved by following the advice “sit less–move more” (indicated by the red arrow), but by increasing sedentary time and decreasing time spent actively and decreasing sleep time (indicated by the black arrow).
It should be noted that this analysis merely serves as a simplified example of how to test the “Sweet-Spot Hypothesis” and it has several limitations. Firstly, given the exploratory nature, our example is based on cross-sectional analysis, which can be subject to biases. Secondly, given the somewhat small sample size, we only had statistical power to control for a few selected potential confounders. Finally, to simplify this example, we decided to combine all physical activities into one variable (i.e., “active”) but it should be acknowledged that each of these activities (i.e., standing, walking, running, cycling and stair climbing) might influence health differentially. However, the proposed approach is applicable to all study designs, and we encourage researchers to test our hypothesis in longitudinal studies, take relevant confounders into account and if possible, consider all 24-h behaviors.