A time compositional analysis of the association between movement behaviors and indicators of mental health in young adults

Movement behaviors (i.e., physical activity [PA], sedentary behaviors [SB], sleep) relate to mental health. Although movement behaviors are often analyzed as distinct entities, they are in fact highly inter‐dependent (e.g., if an individual increases sleep, then PA and/or SB must be reduced) and these dependencies should be accounted for in the analysis. We tested whether perceptions of time spent in movement behaviors (i.e., moderate‐to‐vigorous intensity PA [MVPA], light physical activity [LPA], SB, and sleep) related to depressive symptoms and self‐report mental health in young adults using a compositional analysis. We then estimated change in depressive symptoms with reallocation of time across movement behaviors using compositional time‐reallocation models.

are at higher risk of mental illness later in life and more than 50% who have an episode of major depression experience a recurrence. 7,8hysical activity (PA) is associated with a reduced risk of depression, [9][10][11] and individuals who are more physically active self-report better levels of mental health. 12,13A is a non-pharmacological strategy that may prevent mental health challenges such as experiences of depressive symptoms, and in addition, it supports good mental health.For example, longer more intense PA (e.g., moderate-to-vigorous intensity PA (MVPA)) is associated with lower levels of depressive symptoms relative to light intensity PA. 14 However, PA is only one of several movement behaviors that relates to mental health. 15edentary behavior (SB) has also been associated with depression such that individuals who spend higher amounts of time in SB are more likely to report higher levels of depressive symptoms. 16In addition, sleep relates to depression.Compared to shorter or longer sleep durations, sleep duration between 6 and 8 h daily is associated with better emotional regulation and fewer depressive symptoms. 17While these movement behaviors have been studied in isolation regarding their associations with mental health, understanding how young adults spend time across these movement behaviors may have potential in terms of informing recommendations that aim to prevent or reduce depressive symptoms and support mental health.
The current Canadian 24-h movement guidelines recommend that adults should obtain at least 150 min of MVPA per week (i.e., 21.4 min per day), less than 8 h of SB per day, and 7-9 h of sleep per 24-h period. 18The guidelines also recommend several hours of light physical activity (LPA) each day.While these behaviors may often be considered distinct, considering each movement behavior as distinct obviates that movement behaviors are inherently co-dependent.Specifically, increasing the amount of time spent in PA also affects time spent in SB, sleep, or both.As such, researchers have started to view movement behaviors as part of a "time composition". 15,19Findings from time composition studies to date suggest that the relative time spent in each movement behavior relates to differences in mental wellbeing (i.e., mental health assets indicative of positive attributes such as flourishing), 20 body composition, 15 mortality, 19 and depressive symptoms in children and adolescents. 21,22owever, the extent to which the time composition of movement behaviors relates to depressive symptoms and self-rated mental health in young adults is unknown.
Studies examining movement behaviors as a time composition typically use accelerometers to measure PA, sleep, and sedentary behavior. 19,23However, the amount of time that individuals perceive or self-report that they spend in these behaviors is consequential to health outcomes.5][26] Thus, although selfreport subjective perceptions may be less precise than accelerometer data, there is value in understanding individuals' subjective perceptions of their time spent in movement behaviors.
The purpose of the current study was to examine whether relative time spent in LPA, MVPA, SB and sleep relates to depressive symptoms and to self-rated mental health in young adults using a compositional analysis.Specifically, we explore whether young adults who spend more time in each movement behavior, relative to the other movement behaviors, report differences in depressive symptoms, and self-rated mental health.Further, we explore whether removing or adding 15 min to each behavior, relates to changes in depressive symptoms, or in self-rated mental health.

| Participants and procedure
Data were drawn from the Nicotine Dependence in Teens (NDIT) Study, an ongoing longitudinal investigation of 1294 grade seven students recruited in 1999-2000 in ten Montreal-area high schools. 27Participants were assessed in 20 data collection cycles during the 5 years in high school, and in four cycles post-high school (i.e., cycles 21-24) at mean ages of 20.4,24.0, 30.6, and 33.6.Because the current study was focused on young adulthood, we used data on movement behaviors and indicators of mental health collected in cycle 21 (2007-2008), at age 20.NDIT received ethics approval from Centre de recherche du Centre Hospitalier de l'Université de Montréal (CRCHUM) (ND06.087).

| Movement behaviors
LPA and MVPA were measured using the short version of the International Physical Activity Questionnaire: IPAQ-SF. 28IPAQ-SF is a 6-item questionnaire that measures time spent walking, in moderate intensity PA, and in vigorous intensity PA over the last 7 days.The IPAQ-SF has demonstrated acceptable reliability and criterion validity. 28articipants answered items assessing how many days and minutes per day they spent walking (i.e., a proxy for LPA) doing moderate PA, and doing vigorous PA, over the last 7 days.Average minutes spent daily walking and in each of moderate and vigorous intensity PA were then calculated by multiplying the number of days by the average number of minutes each day, and then dividing this score by seven to obtain average daily minutes.As per the IPAQ-SF data processing guide (www.ipaq.ki.se, 2005), activity bouts greater than 180 min daily were windsorized to allow a maximum of 1260 min per week in each PA category (i.e., vigorous intensity, moderate intensity, and walking).Average minutes in vigorous and moderate intensity PA were summed to obtain average time spent in MVPA daily.Time spent in both LPA and MVPA was treated as continuous variables.
SB was measured in cycle 21 using four items adapted from a systematic review, 29 including: "How many hours of television (including video movies) do you usually watch in a single day?"; "How many hours do you usually spend on a computer in a single day for school or at work?"; "How many hours do you usually spend on a computer in a single day during your leisure time (playing computer games, using the Internet)?; and "How many hours do you usually spend reading (books, magazines, newspapers, homework) in a single day?Participants recorded the number of hours spent in each behaviour on a usual weekday and on a usual weekend day; they were instructed to: "Write "0" if none; and Write "LT ½" if less than ½ hour."Average time spent daily in SB was then calculated using the following formula: [(sum of hours per a weekday × 5) + (sum of hours per a weekend day × 2)]/7.Time calculated in hours was multiplied by 60 to obtain the average number of minutes spent in SB daily.In this analysis, time spent in SB daily was treated as a continuous variable.
Sleep duration was measured using two items adapted from the Pittsburgh Sleep Quality Index (PSQI) 30 : "In the past month, at what time did you usually go to sleep at night?" and "In the past month, at what time did you usually wake up in the morning?".Participants reported both go-to-sleep time and wake-up time in a 24-h format.In this analysis, time spent in sleep daily was treated as a continuous variable.

| Outcome variables
Depressive symptoms were measured using the Major Depression Inventory (MDI), a 10-item measure in which individuals self-report experiences of depressive symptoms over the previous 2 weeks. 31Response options for each item were recorded on a 6-point Likert-type scale ranging from 0 (At no time) to 5 (All the time).MDI scores range between 0 and 50, with higher scores denoting a higher likelihood of a probable case of depression.The MDI has demonstrated validity and high internal consistency in an earlier study (α = 0.94) 31 and in NDIT cycle 21 (α = 0.88).In this analysis, the MDI score (i.e., depression symptoms) was treated as a continuous variable.
Self-rated mental health was assessed in one item "How would you rate your mental health?"Participants responded on a scale from 1 (poor) to 5 (excellent).This item was used in the Canadian Community Health Survey 32 and is commonly used in large-scale epidemiologic studies and surveillance surveys. 33,34Further, this type of single item self-report measure predicts mortality over and above many objective indicators of health such as physician examinations and medical records. 35,36Overall, older adults who perceive their health as excellent are six times less likely to die compared to older adults who perceive their health as poor. 37As such, self-report subjective perceptions of mental health are valuable tools in understanding factors related to mental health.This item was treated as continuous.

| Covariates
Consistent with previous research examining movement behaviors and health outcomes, 15,19 age, sex, and highest education level attained were included in the analysis to account for known associations between those variables and health outcomes.Education level was categorized into 4 levels, (1) Did not graduate high school, (2) High school graduate, (3) Technical school (even if not graduated), ( 4) University (even if not graduated).

| Analysis
A compositional analysis using data collected in a crosssectional study design, was used to examine associations between movement behaviors (LPA, MVPA, SB, and sleep) and mental health.The compositional geometric mean is a measure of central tendency and was computed as the geometric mean of time spent in a movement behavior after being normalized to the proportion of total time.All data analyses were conducted using R version 4.2.1. 38A total of 786 of the 1294 participants who participated in the NDIT study had complete data on all study variables at Cycle 21.Of the 786, 16 participants reported more than 20 h per day in a single movement behavior (e.g., sleep) and were excluded because these values are highly improbable.Thus 770 participants were retained for analysis.
The predict_delta_comps function within the codaredistlm package in R 39 was used to conduct isometric log ratio (ilr) multiple linear regression models.Ratios of movement behaviors in the simplex space were transformed into equivalent isometric log-ratio (ilr) coordinates in real space that can be used in standard regression models.Log-ratios cannot be applied to zero values.Therefore, zero values in the MVPA and LPA variables were replaced by 0.93, which represents 65% of the detection limit in the IPAQ-SF, at 10 min per week or 1.42 min per day. 40Compositional multiple regression models that use the ilr coordinates of movement behaviors were fitted to examine the association between time spent in MVPA, LPA, SB, and sleep, and each of the outcome variables controlling for age, sex, and education.Models were fit sequentially so that each behavior played the role of the first part of the composition.Specifically, we included the coefficient of the first ilr coordinate of each behavior in each model to describe the association between the proportion of time spent in this behavior, relative to other behaviors, and the outcome variables. 41Next, a compositional isotemporal reallocation model was used to estimate the effect of displacing 15 min of time from one movement behavior to another movement behavior, while holding the other movement behaviors constant. 41While 30 min is the most common reallocation time, 42 Corder and colleagues indicate that 10 min should be the minimal starting point when replacing SB with PA. 43 We examined a reallocation time of 15 min as this is likely the maximum amount of time people could be realistically expected to increase MVPA in a day.

| RESULTS
Descriptive statistics and correlations between study variables are available in Table 1.The average age of participants was 20.4 years and 55% were female.Of the 770 participants, 7.4% did not graduate from high school; 11.4% graduated from high school; 61.8% attended and/ or graduated technical school, and 19.3% attended and/ or graduated from university.Participants were younger in the analytic sample (M = 20.4,SD = 0.71) compared to those who did not participate or were not retained for analysis (M = 20.6,SD = 0.93).There were 343 males and 427 females in the analytic sample.Of those who did not participate or were not retained, 59 were male and 46 were female.Figure 1 details a participant flow chart outlining how the final sample was derived.Overall, the geometric means suggest that participants spent 57% of their time in sleep, 40% of their time sedentary, 2% of their time in LPA, and 1% of their time in MVPA.In accordance with MDI scoring, 44 696 individuals reported scores consistent with "no or doubtful depression", 37 reported mild depression, 18 reported moderate depression, and 19 reported severe depression.
In a multiple linear regression model controlling for covariates, movement behaviors were not significantly associated with depression, F(761, 3) = 2.49, p = 0.06, and significantly associated with mental health, F(760, 3) = 5.13, p < 0.01.Controlling for covariates and holding all movement behaviors at their geometric mean, the predicted T A B L E 1 Means, standard deviations, and correlations with confidence intervals.The isotemporal substitution models exploring the effect of reallocating 15 min from one movement behavior to another on the mental health indicators are presented in Table 3 and illustrated in Figure 2. The most notable changes in the mental health indicators were observed when adding to or subtracting time from MVPA.For example, subtracting 15 min from MVPA and adding it to sleep was associated with a 0.46 (95% CI = −0.02,0.90) increase in depressive symptoms, although this was not statistically significant.Further, subtracting 15 min from MVPA and adding it to LPA was associated with a significant 0.57 (95% CI = 0.03, 1.08) increase in depressive symptoms.Adding 15 min to MVPA from SB was associated with a significant −0.18 (95% CI = −0.36,−0.01) reduction in depressive symptoms.Similar trends in self-rated mental health were observed after adding 15 min from LPA, SB, or sleep to MVPA (Table 3).

| Sensitivity analysis
When all movement behaviors are aggregated, on average 945 min within a total day was accounted for (SD = 236).This varies from the total of 1440 min in a day.A sensitivity analysis was conducted in which we included only individuals who reported their total time in movement behaviors ±236 min (±1 SD) of the true 1440-minute day (n = 95).Overall, geometric means indicated that individuals who were within 236 min reported less time in sleep (40%) and more time in sedentary behavior (57%), which is consistent with studies measuring 24-h movements using accelerometer data. 19Similar to the main analysis, light and moderate to vigorous PA represented 2% and 1% of an individual's total day, respectively.In composition analyses, trends among those who were within ±236 min of the 1440-minute day were similar to the overall sample.Specifically, time spent in MVPA relative to other movement behaviors was nonsignificantly and negatively associated with depressive symptoms, β = −0.63,SE = 0.53, p = 0.24, and significantly positively associated with self-rated mental health, = 0.13, SE = 0.06, p = 0.04.As in the main analysis, time spent in sleep, SB, and LPA relative to other behaviors, were not significantly associated with depressive symptoms or mental health.Further, sensitivity analysis examining associations with individuals using antidepressant medication removed indicated similar results.To maximize sample size, these individuals were included in the analysis.

| DISCUSSION
This study examined whether relative time spent in LPA, MVPA, SB, and sleep relates to depressive symptoms and self-rated mental health in young adults.The Note: Models adjusted for age, sex, and education level.
results suggest that young adults who spent more time MVPA relative to other movement behaviors reported fewer depressive symptoms and better self-rated mental health.Further, reallocating 15 min of LPA to MVPA was associated with reductions in depressive symptoms and improvement in self-rated mental health.Similarly, reallocating 15 min of MVPA to LPA, SB or sleep was associated with increases in depressive symptoms and decreases in self-rated mental health.There was no change in the mental health outcomes after adding or subtracting 15 min between LPA, SB, and sleep.These results indicate that self-reported PA, relative to other self-report behaviors, relates to depressive symptoms.This novel finding extends extant 24-h movement composition analyses, which typically use accelerometer data.Specifically, this study highlights the potential to evaluate interventions targeting 24-h movement behavior using self-report data.Thus, reflective of previous studies indicating positive effects of PA on both physical and mental health, 9,10 our findings support that time spent in MVPA relative to other movement behaviors is related to mental health.
Participants reported spending 1% of their day in MVPA on average, which is near the 150 min of MVPA per week recommendation detailed in the Canadian public health guidelines. 18Our data indicate that reallocating more time to MVPA from any other movement behavior is positively associated with improvements in depression symptoms and mental health.Consistent with previous research, the strongest associations were observed when MVPA was displaced by another behavior. 41,45This finding may relate to the portion of time MVPA takes up in ones' day, whereby, removing 15 min of MVPA is a more significant change than removing 15 min of sedentary behavior.This highlights that mental health not only relates to improvements in PA patterns, but also relates to drop offs in MVPA.Further, compared to other behaviors, MVPA requires higher levels of energy expenditure, 46 and often involves participation in the context of a social environment. 47,48Both fitness 49 and social factors 8 are positively associated with mental health.However, the mechanisms explaining why higher levels of MVPA compared to other movement behaviors relate to mental health need further investigation.Overall, these findings highlight the value of including MVPA within a 24-h timeframe, and the potential detrimental impact removing MVPA can have on mental health.
Young adults often experience changes in their living circumstances (e.g., moving out of the family home, beginning university, developing romantic partnerships, establishing careers) as they develop independent lives, which can result in reduced participation in MVPA such as sport. 50Based on the current findings, reductions in MVPA may partly explain the heightened risk of depression in this age group.If future studies confirm that reducing MVPA in favor of other movement behaviors such as SB does indeed affect mental health negatively, public health interventions and messaging targeting mental health in young adults may need to focus on encouraging young adults to maintain or increase MVPA during these life transitions, rather than changing LPA, SB, or sleep.This study included subjective perceptions movement behaviors, rather than device-measured movement behaviors.Although self-report measures are susceptible to retrospective bias, self-report movement behaviors represent intentional behaviors in which individuals are consciously aware they are participating in, which are unique from device-measured PA, sleep, and SB.It is possible that the associations observed herein relate to the psychological aspects pertaining to movement behaviors, 24 which may be precursors to modifiable intervention efforts.As such, beyond encouraging individuals to be more physically active, researchers should consider how reframing relevant activities as PA (e.g., washing the dishes, vacuuming) to enhance potential benefits on mental health.There were several noteworthy limitations in this study.First, the movement behaviors were categorized broadly and there are likely many contextual factors such as participation alone or with others, 51 and different modalities such as type of PA 52 that could influence the strength and direction of associations between movement behaviors and mental health.Likewise, there are many types of sedentary behavior (e.g., eating, socializing, driving) that have differing impacts on mental health, but were not captured within this study.Further research on time spent in different types of movement behaviors may highlight important contextual factors and specifics of engagement.Additionally, associations in this paper are examined in a cross-sectional study design and as such, it is possible participants' mental health impacted their movement behaviors (i.e., the association is bi-directional).Further, recent research indicates that depressive symptoms relate to the time day in which individuals engage in movement be- 53 Future research will need to examine whether time of day is relevant in this domain.Also, PA was measured as an average across a week, sleep was measured as an average across a month, and sedentary behavior was measured across a typical day.These different reference points may have introduced bias into individuals' selfreport, however, because we examine behaviors as a daily average, it is unlikely these differences had a significant impact on the results.Finally, a possible limitation of generalizability relates to the longitudinal data that were collected over a decade ago.Advances in technology and communication, in conjunction with the pandemic, have evolved such that screen time and SB may have increased over the past decade, suggestive that the current results may be conservative estimates.
In conclusion, our results suggest that time spent in MVPA relative to LPA, SB, and sleep is related to mental health.If replicated, public health recommendations aimed at improving or sustaining mental health in young adults may need to focus on increasing or sustaining MVPA by reallocating time spent in other behaviors.

| PERSPECTIVE
Young adulthood is a time of significant transition which can impact mental health. 54Movement behaviors such as PA directly impacts young adults' mental health. 50As such, there is growing attention on developing movement interventions to support mental health.However, the use of accelerometer devices to measure movement behavior is costly and not always practical.The current study highlights the utility of using self-report measures of movement behaviors to understand the efficacy of mental health interventions.Further, this research indicates that movement behaviors, as they relate to mental health, can be understood as a composition, whereby increases in one behavior (e.g., PA) comes at the expense of another (e.g., sleep).Therefore, instead of targeting one behavior at a time, practitioners looking for behavioral strategies to support mental health should consider replacing time spent in sedentary behavior with an increase in moderate to vigorous PA.

F I G U R E 2
Changes in depression and mental health from adding and removing 5, 10, and 15 min of moderate to vigorous physical activity (MVPA), light physical activity (LPA), sedentary behavior (SB), and sleep.

in depressive symptoms with reallocation of 15 min…
Change in prediction matrix showing changes in depressive symptoms and mental health, with reallocation of 15 min from the behaviors in the columns to the behaviors in the rows (n = 770), NDIT 2012.Values in brackets are 95% confidence interval.Negative values indicate reduction in predicted depressive symptoms and mental health values with time reallocation, while positive values indicate an increase in predicted values with time reallocation.