Article Text
PDF
Abstract
Background Sedentary behaviour (SB) is detrimental to cardiometabolic disease (CMD) risk, which can begin in young adulthood. To devise effective SB-CMD interventions in young adults, it is important to understand which context-specific SB (CS-SB) are most detrimental for CMD risk, the lifestyle behaviours that cluster with CS-SBs and the socioecological predictors of CS-SB.
Methods and analysis This longitudinal observational study will recruit 500 college-aged (18–24 years) individuals. Two laboratory visits will occur, spaced 12 months apart, where a novel composite CMD risk score (eg, arterial stiffness, metabolic and inflammatory biomarkers, heart rate variability and body fat distribution) will be calculated, and questionnaires to measure lifestyle behaviours and levels of the socioecological model will be administered. After each laboratory visit, total SB (activPAL) and CS-SB (television, transportation, academic/occupational, leisure computer, ‘other’; ecological momentary assessment) will be measured across 7 days.
Ethics and dissemination This study has received full ethical approval, and participants provide written informed consent. Our hypothesis is that certain CS-SB will show stronger associations with CMD risk, compared with total sedentary behaviour (T-SB), even after accounting for coexisting lifestyle behaviours. We also expect a range of intra-individual, inter-individual and physical environmental socioecological factors will predict CS-SB. Findings addressing both the primary and any secondary research aims will be submitted for publication in a high-impact peer-reviewed journal.
- Behavior
- Cardiovascular Disease
- Physiology
- Observational Study
- Physical Fitness
- PUBLIC HEALTH
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
This study proposes a novel measure of context-specific sedentary behaviour (SB) derived from merging ecological momentary assessment (EMA) and accelerometry data.
This study employs a data-driven approach to calculate an original cardiometabolic disease risk score for college-based young adults, incorporating novel and traditional biomarkers.
The study applies the socioecological framework to identify predictors of context-specific SB across intrapersonal, interpersonal environmental levels.
Although well-suited for addressing the aims of this study, our EMA protocol may not fully capture all context-specific SB throughout the 24-hour day.
Our findings will have limited generalisability beyond college-based young adults (eg, older adults, office workers).
Introduction
Cardiometabolic disease (CMD) is a critical health burden in young adults, with approximately 10–20% of this population exhibiting advanced but asymptomatic atherosclerotic lesions.1–3 CMD trajectories in this population are also worsening, with type 2 diabetes prevalence increasing from 0.9% (1971–75) to 3.2% (2009–2012) among adults aged 20–44 years, exceeding the trajectory in those older than 45 years.4 In the USA, two-thirds of young adults attend college,5 6 where they often adopt poor lifestyle behaviors7 8 that can track into older age and contribute to CMD development.9 10 Consequently, young adulthood is an important, yet under-researched period to implement lifestyle behaviour interventions that aid the primary prevention of CMD.
The definition of sedentary behaviour (SB) is any low intensity activities (<1.5 metabolic equivalents) performed while seated, reclined or supine.11 SB is biologically distinct from physical inactivity (defined as not meeting physical activity guidelines), such that individuals can be both highly sedentary and highly physically active.12 In college-based young adults (CBYA), SB has received considerably less research attention compared other lifestyle behaviors.13 14 This is despite SB being the dominant waking behaviour in this population, with accelerometry-derived estimates of 10–11 hours/day.15 16 Worryingly, a moderate-to-strong association exists between total SB time and CMD risk,16 even in those meeting moderate-to-vigorous physical activity (MVPA) guidelines.17 18 Consequently, public health agencies, including the WHO, have called on the research community to better understand how to implement SB-reduction interventions.14 19
To develop effective SB-CMD interventions, it is necessary to fully elucidate the link between SB and CMD, but this relationship is poorly understood in CBYA. This is likely due to the multidimensional complexity of SB, which occurs across a wide spectrum of domains, including television viewing, occupational sitting, seated transport, and leisure-time computer use and interindividual (eg, alone, with friends) and environmental (at home, at work) contexts. Cross-sectional research in general adult populations using retrospective self-report measures indicates that television viewing is more strongly associated with CMD risk relative to other CS-SBs.19 The reasons for this are not well understood, but may include the simultaneous engagement with unhealthy behaviours, such as the (over-)consumption of processed foods.20 However, whether total or CS-SB is more strongly associated with CMD risk, and how coexisting lifestyle behaviours influence this association, is not known in CBYA.
We also have scant understanding of the variables that predict CS-SB in CBYA, but this information is critical for the development of successful SB interventions. The socioecological model (SEM) has been used with good effect to increase MVPA levels21 and highlights that SB interventions are unlikely to be successful unless multilevel predictors of SB are considered, including intrapersonal (eg, awareness of time spent engaging in SB), interpersonal (eg, friends involvement with SB) and physical environmental (eg, living environment) factors.21–23 One longitudinal study found that social-cognitive variables like attitudes, self-efficacy and social norms were the strongest predictors of CS-SB, while physical environment factors showed limited predictive value.24 However, this line of research has solely been conducted in general adult populations and has predominantly focused on non-modifiable socio-demographic variables,23 25 which ignores the full spectrum of SEM factors.26
In summary, CMD is a growing concern in CBYA, and therefore it is critical to develop effective interventions and public health recommendations for this population. However, in CBYA, we do not yet fully understand the link between CS-SB and CMD risk, the influence of behaviours that co-exist with CS-SB on CMD risk and the socioecological predictors of CS-SB. Without this information, it is not possible to devise highly effective SB interventions that are capable of mitigating CMD risk. Therefore, the aims of the Cardiometabolic Outcome Negation Through Early-adulthood ConteXT-specific Sedentary Behavior reduction (CONTEXT-SB) study are:
Aim 1. To identify whether total sedentary behaviour (T-SB) or CS-SB is more strongly associated with CMD risk in CBYA.
Aim 2. To determine whether the Aim 1 CMD risk is directly explained by SB or mediated by co-occurring lifestyle behaviours (eg, diet, physical activity) that cluster with T-SB and CS-SB.
Aim 3. To investigate which SEM factors (intra-individual, inter-individual, physical environment) are associated with T-SB and CS-SB.
The overall deliverable (underpinned by the three aims above) of CONTEXT-SB is to develop an evidence-based, multilevel intervention to target SB reduction in CBYA.
Methods
CONTEXT-SB is a longitudinal observational study conducted within the Cardiometabolic Laboratory (CML) at the University of North Carolina at Chapel Hill (UNC-CH). This study commenced data collection in December 2022 and is anticipated to be completed in May 2026. There are two identical laboratory visits, each lasting ~120 min, separated 12 months apart (figure 1). After the completion of each laboratory visit, a 7-day movement behaviour assessment period occurs, which includes measuring CS-SB and T-SB. We measured cardiometabolic biomarkers 12 months apart to align with other research24 and to balance participant burden, retention and budgetary constraints. The longitudinal component of the study will also allow us to assess the stability of our novel CMD risk score over time.
Visualisation of the CONTEXT-SB study protocol. After enrolling in the study, participants are scheduled for their visit 1 assessments at the Cardiometabolic Lab, where cardiometabolic disease risk is established and movement behaviour devices are deployed. The visit 2 protocol is identical to visit 1. bfPWV, brachial-femoral pulse wave velocity; cfPWV, carotid-femoral pulse wave velocity; CONTEXT-SB, Cardiometabolic Outcome Negation Through Early-adulthood ConteXT-specific Sedentary Behavior reduction; HRV, heart rate variability; NWT, non-wear time; PA, physical activity; PWA, pulse wave analysis; PWV, pulse wave velocity; SB, sedentary behaviour; SEM, socioecological model.
Ethics and dissemination
The study has received full ethical approval from the University of North Carolina at Chapel Hill Institutional Review Board (IRB#: 22–0819), and participants provide written informed consent at the start of the first laboratory visit. The publication of this protocol paper marks the initial phase of disseminating findings from the CONTEXT-SB study. We anticipate that preliminary and methodological findings will be published in early 2025, while primary findings for aim 1, aim 2 and aim 3 will be shared in late 2026.
Patient and public involvement
Neither the public nor participants were involved in the development of the CONTEXT-SB protocol. Participants can receive a feedback report outlining their data (eg, blood pressure levels), and the findings of this study will be disseminated to the public through appropriate open-access research journals (where appropriate), conferences and social media platforms.
Inclusion criteria
We aim to recruit 500 CBYA from UNC-CH. Inclusion criteria are: (i) college students aged 18–24 years, (ii) available for follow-up testing 12 months after the first lab visit and (iii) plan to be on campus at least 1 month prior to the follow-up laboratory visit. An inclusive recruitment strategy was adopted to better reflect the diverse health and behavioural characteristics of CBYA. Before each laboratory visit, the following pre-assessment guidelines are followed: fasted for 12 hours, no alcohol consumption for 24 hours, no consumption of supplements for 24 hours and no strenuous/vigorous exercise for 24 hours.
To ensure the sample is sociodemographically (eg, age, sex, race) representative of the UNC student body, a quota sample approach is used. Our primary recruitment strategies are classroom visits, flyers, mass emails and a UNC-CH participant recruitment website. To boost recruitment and retention, flexible scheduling (weekends) and automated follow-up reminder messages via the Calendly software27 are used. Participants receive a $50 gift card for each study visit and are entered into a random prize draw for an Apple Watch. It is conservatively estimated that 80% of participants (n=400) will complete both study visits.
Assessment visits
On arrival at the CML, pre-assessment guideline adherence is confirmed. Acquisition of consent (DocuSign) is completed, and height (stadiometer; Perspective Enterprises; Portage, MI) and body mass (Newell Brands; Atlanta, GA) are measured. Self-reported information on a range of sociodemographic factors, such as race, ethnicity, highest level of education acquired, current year of study, college major, parental occupation, age and any regular medications, are recorded using standard questionnaires.
Completion of each aspect of the study protocol is recorded via electronic data capture software (REDCap).28 29 REDCap is a secure, web-based database designed to support data capture for research studies, providing (1) an intuitive interface for validated data capture; (2) audit trails for tracking data manipulation and export procedures; (3) automated export procedures for data downloads to common statistical packages; and (4) procedures for data integration and interoperability with external sources. A full visualisation of the study protocol is provided in figure 1.
Primary outcome: composite cardiometabolic disease risk score
The primary outcome measure for CONTEXT-SB is a composite CMD risk score, which is comprised of 14 novel and traditional cardiometabolic markers (see table 1).
Outcomes (n=14) that form the composite cardiometabolic disease risk score
Cardiovascular outcomes
The VICORDER system (SMT Medical Technology GmbH, Wuerzburg, Germany) measures carotid-femoral pulse wave velocity (cfPWV) and brachial-femoral pulse wave velocity (bfPWV)), which are markers of arterial stiffness. There are 15 min of quiet rest before the first measurement, and all measures are taken in triplicate with 1 min of quiet rest in between; the closest two being averaged for later analysis. In accordance with the manufacturer’s guidance, participants are positioned in a semirecumbent posture at an incline of 25°. This positioning helps to minimise interference from the jugular vein in the detection of the foot of the carotid upstroke. Briefly, the device calculates pulse transit time via oscillometry at two arterial sites, and manufacturer guidelines were followed for capturing the straight-line distances between measurement sites. For measures of cfPWV, the straight-line distance from the carotid artery to the femoral artery is multiplied by a factor of 0.8, to more accurately reflect the actual physiological path distance.30 Additionally, the VICORDER system calculates oscillometric resting blood pressure and additional outcomes from pulse wave analysis, including augmentation index, central systolic blood pressure and mean arterial pressure.
The MindWare Mobile system (MindWare Technologies, Ltd., Westerville, OH) receives signals from 3-lead (right clavicle, lower left and right rib) ECG which are processed continuously using BioLab software. Data are collected for the last 5 min of the 15-min resting period in a semirecumbent position. Guidelines for quality control are followed, including cleaning electrode sites with 70% ethanol prior to electrode placement.31 32 A visual assessment of high-quality signal is confirmed before data collection is initiated. The primary heart rate variability (HRV) outcome is the root mean square of the successive difference. Secondary outcomes from HRV may include non-linear HRV metrics (eg, entropy) and frequency-domain parameters such very-low frequency HRV (0.00–0.04 Hz), low frequency HRV (0.04–0.15 Hz) and high frequency HRV (0.15–0.40 Hz).33
Body fat distribution
Full body dual-energy X-ray absorptiometry scans (Hologic, Horizon; Bedford, MA) are performed to assess total body fat and trunk fat percentage. Participants are instructed to wear clothing free from metal and to remove accessories prior to the scan. Participants are positioned supine according to manufacturer guidelines with arms placed to the side, palms pronated and legs internally rotated such that the toes are touching. If a participant’s height exceeds the examination area, the lower limbs are prioritised, as the head composition is estimated.
Metabolic and inflammatory outcomes
An aseptic venous blood draw is performed for the collection of two 10-mL whole blood samples within serum and ethylenediamine tetra-acetic acid (EDTA (K2)) tubes (BD Vacutainers, Becton, Dickinson and Company, Franklin Lakes, NJ, USA). Immediately postcollection, 1.5 µL of EDTA whole blood is used for assessing glycated haemoglobin via a multiassay analysis platform (AFINION 2 Analyzer, Abbott Diagnostics Technologies AS, Oslo, Norway), while the serum sample is allowed to clot at room temperature for 1 hour. Both samples are centrifuged at room temperature for 15 min at 1200×g. A 40-µL sample of the serum supernatant is used to generate a haematological metabolic profile (eg, triglycerides, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, fasting blood glucose) via reflectance photometry (Cholestech LDX Analyzer, Abbott Diagnostics Technologies AS, Oslo, Norway). Remaining supernatants are aliquoted and stored at –80°C until future analysis. Planned outcomes include insulin and C-reactive protein, which will be assessed via ELISA according to manufacturer instructions.
Aim 1 exposure variables
The aim 1 exposure variables are T-SB and five CS-SBs.
Context-specific sedentary behaviours
The Smartphone Ecological Momentary Assessment (SEMA3; Melbourne eResearch Group, Melbourne, Australia) application is installed on each participant’s personal mobile device.34 Participants are promoted to complete nine daily surveys, with one survey programmed to be distributed during each of these blocks: 00:00–02:15, 02:45–05:00, 05:30–07:45, 08:15–10:15, 10:45–12:45, 13:15–15:15, 15:45–18:00, 18:30–20:45 and 21:15–23:30. This approach was selected to account for the entire 24-hour day, and the varied sleeping patterns of CBYA.35 The surveys measure a range of intra-individual, inter-individual and environmental factors, including CB-SB (occupation/study; leisure computer; screen time; transportation; other—see figure 2). Participants are instructed to reflect on their behaviour immediately prior to receiving the survey and to complete the surveys as quickly as possible, as each survey becomes unavailable 30-min postdeployment. Participants are alerted via the SEMA3 application if they fall below a set compliance threshold of 60%.
Ecological momentary assessment (EMA) sequence. Participants are asked to reflect on the context-specific sedentary behaviours they were engaging in at the moment they received the survey.
Total sedentary behaviour
The activPAL (PAL Technologies Ltd.; Glasgow, UK) is a gold-standard inclinometer which records data at 20 Hz (2 s) and is affixed on the anterior portion of the right thigh (midpoint between the anterior superior iliac spine bony landmark and knee bone) using a nitrile/latex sleeve with waterproof Tegaderm dressing. Participants are instructed to remove the device only for high-impact contact sports or during swimming in deep environments where it may be lost. The activPAL captures sedentary behaviour (T-SB) data for seven full days (24 h/day), beginning at 11:59pm the day of each study visit, using established methods.36 37 The PALanalysis software will be used to calculate T-SB and other SB outcomes using the CREA algorithm.36–38 Following National Health and Nutrition Examination Survey protocols,39 40 a valid day of activPAL monitoring is defined as at least 16 hours of wear time, with three valid weekdays and one valid weekend day being required for data inclusion.
An automated survey is sent to participants via REDCap each morning of the 7-day movement behaviour assessment period. Participants report their sleep, wake and nap times. This information is collected to cross-validate sleep data from the objective sleep device (see below) and activPAL. Participants also report any time periods in which the activPAL was removed, to quality check the non-wear times estimated by the activPAL.
For a small subset (n=20) at the start of the study, the MOX1 accelerometer (Maastricht Instruments BV; Maastricht, NL) was used. However, a lack of functionality to detect non-wear and sleep periods was found to be prohibitive for continued use in the study. To ensure within-subject data comparability, the same accelerometer will be used for the same participant across visits 1 and 2.
Combination of ecological momentary assessment (EMA) and activPAL data
Using a custom R script, the activPAL epoch-level data will be aggregated into 30-min segments (eg, 01:00–01:30, 10:30–11:00, 16:00–16:30). The timestamped ecological momentary assessment (EMA) survey responses will be merged with the activPAL data, assigned to the appropriate 30-min time block (eg, a survey completed at 14:15 will be assigned to the 14:00–14:30 block). These data will then be combined to calculate time spent (per week) in each CS-SB. To offer an example, a participant may engage in 60 hours of activPAL-measured T-SB throughout the 7-day observation period. The EMA data shows that a participant responded to 52 out of the 63 possible surveys (nine surveys per 7 days of observation). Of these 52 responses, 40 were reported to be SB, and 30 of these 40 SB responses were identified to be in the occupation/study domain. Thus, time spent in occupational/study-based SB is calculated as 45 hours across the week or 75% (30/40*100).
Aim 2 exposure variables
Below (and see table 2) are the 11 lifestyle behaviour exposure variables for aim 2, which may partially explain associations between T/CS-SB and CMD risk.
Lifestyle behaviour variables (n=11) for aim 2 of the CONTEXT-SB study
Physical activity
Time spent engaging with light physical activity and MPVA will be derived from the raw acceleration data acquired from the activPAL and MOX1 accelerometers (see above). These data will be derived in R using the GGIR package.41 42 Established cut points for raw acceleration data from the thigh will be applied in analyses.43 Validation criteria for valid days is described above in the activPAL SB section.
Diet
Dietary patterns will be derived from the Diet Health Questionnaire III (154 items).44 Specifically, we expect principal component analysis to generate three dietary patterns: processed foods, fruit and vegetable consumption, and breakfast foods.
Substance use
The Alcohol, Smoking, and Substance Involvement Screening Test (12 items, intra-class correlation (ICC)=0.90 to 0.97)45 will measure specific involvement scores for 12 different substances (tobacco, alcohol, cannabis, cocaine, prescription stimulants, methamphetamines, inhalants, sedatives or sleeping pills, hallucinogens, heroin, prescription opioids and other substances). For the primary analysis, we will derive three variables which are considered most specific to our CBYA population: a score for tobacco and alcohol and a global risk score which sums together all drug classes.
Sleep
Objective sleep data will be captured using the SleepScore Max device (Sleep Solutions LLC, Carlsbad, CA), linked to the SleepScore application downloaded on each participant’s personal mobile device. In line with manufacturer guidelines, participants are instructed to place the device at chest level on a surface next to their bed for all sleeping and napping periods across the 7-day observation period. The device uses sonar technology to non-invasively track sleep without coming into contact with the participant and potentially interfering with natural sleeping patterns. The SleepScore Max device has been validated against polysomnography (sensitivity=0.94, accuracy=0.88).46 Participants are instructed to activate the SleepScore Max device immediately before they begin sleeping and to end the sleep tracking immediately when they awake from each bout of sleep. The 5-item Reduced Morning-Eveningness Questionnaire47 is administered to examine the sleep chronotype of each participant (eg, morning-oriented or evening-oriented).
Aim 3 exposure variables
Sixteen (SEM) factors will form our selection of aim 3 exposure variables (see table 3). These were selected with consideration to theoretical models, previous population use and brevity (88 total items).
Socioecological model variables (n=16) included for aim 3 of the CONTEXT-SB study
SEM predictors of sedentary behaviour Cronbach’s ⍺ are derived from previous literature. The same four variables for the physical environment questionnaire are presented for the college-environment and home-environment separately.
Intra-individual
Five intra-individual variables will be measured (see table 3). Self-efficacy will be assessed using a modified version of the Physical Activity Self-Efficacy scale.48 Self-regulation, outcome expectations and personal barriers will be evaluated through a modified Cognitive Behavioral Physical Activity Questionnaire.49 Additionally, psychological stress will be measured using the Perceived Stress Scale50 and the Personal Burnout Scale.51
Inter-individual
Two inter-individual variables will be measured: social norms and social support. Social norms will be evaluated by adapting the scale used by Ball et al.52 Participants will respond to the following statements using a 5-point Likert scale: (i) ‘I often see other people purposefully interrupting their SB’; (ii) ‘Lots of people I know purposefully interrupt their SB’; and (iii) ‘Lots of people I know regularly engage in SB for long periods without interruption’. Social support will be measured by adapting the scale used by Sallis et al.53 Using a 5-point Likert scale, respondents will rate the frequency with which friends or colleagues, during the past year: (i) purposefully interrupted my SB, (ii) encouraged me to reduce SB and (iii) discouraged me from excessive SB.
Physical environment
Four physical environment variables will assess participants’ perceptions of the college environment and their current residence, regarding: (i) functionality, (ii) safety, (iii) aesthetics and (iv) destinations. These variables will be evaluated using the Physical Environmental Neighborhood Factors scale.54
Covariates
Additional standard surveys to collect sociodemographic (eg, age, sex, race, ethnicity) information will also be administered, as will the International Standard Classification of Occupations survey (ISCO-88) to classify parental occupation (eg, manager, professional, service and sales worker) based on the primary occupation of the adult in the family household who earns the most income (ie, primary breadwinner).55
Statistical methods and sample size consideration
The primary outcome is a composite CMD risk score derived using factor analysis from the outcomes shown in table 1.56 The number of factors will be determined using parallel analysis and cross-validation. The factors will be subject to orthogonal ‘varimax’ rotation. If varimax rotation fails to produce interpretable factors, non-orthogonal rotations will be implemented. We will use an a priori loading greater than 0.40 to interpret the factor pattern. We will derive a CMD risk score from this factor structure by summing the individual factor scores.56
Aim 1 analysis
Associations between CMD risk with T-SB and each of the five CS-SB (see figure 2) will be estimated using linear mixed-effects models. An exchangeable working covariance structure will account for repeated measures within the same individual over time. Adjustments will be made for multiple comparisons using either Bonferroni or Benjamini-Hochberg correction. The strongest predictor of CMD risk will be identified as the largest standardised coefficient.
Aim 2 analysis
A two-stage approach will be used.57 First, we will conduct agglomerative hierarchical cluster analyses based on the Ward method with squared Euclidean distance to identify the potential number of clusters among participants and calculate the initial centroid values for each cluster. Second, we will perform a k-means cluster analysis using the number of clusters identified in stage one to form the final clustering result. To examine the stability of the final clustering, we will randomly divide the total sample into two halves, and the described process will be repeated. Cohen’s k coefficients will be calculated to measure the degree of agreement between the classification of participants using the total sample and each half-subsample.
Longitudinal mediation effects will be examined using the two-step longitudinal parallel process (LPP).58 In the first step, the intercept and slope of each longitudinal variable will be estimated separately for each individual using multilevel modelling. This approach will account for any missing longitudinal data. In the second step, separate structural equation models will test for the longitudinal mediation effects of the lifestyle clustering behaviours (table 2) on the relationship between T/CS-SB and CMD risk. Additionally, if the data are non-normally distributed, bootstrap resampling will be implemented to obtain more stable and valid estimates of the standard errors.59 The final model will be adjusted for age, sex, race and parental occupation.
Aim 3 analysis
Associations between T/CS-SB and each SEM variable (table 3) will be evaluated using separate linear mixed-effects models. An exchangeable working covariance structure will account for repeated measures within the same individual over time. The SEM variables will be specified separately, then collectively, using a stepwise model. The final model will be adjusted for age, sex, race and parental occupation. While not a primary aim, in a secondary analysis, we will additionally test sex and race as potential moderators and conduct stratified analysis if appropriate.
Sample size justification
The recruitment target for CONTEXT-SB is n=500 individuals, which we anticipate will result in a total of 900 observations—assuming two observations per participant with 80% retention for visit 2.
Aim 1. Two different approaches are commonly used to estimate sample size in factor analysis: (i) a minimum participant-to-item ratio ranging from 5:1 to 10:160 61 or (ii) a minimum number of participants ranging from 50 to 400.62–64 We opted to satisfy the very good to excellent criteria suggested by Comrey and Lee, where: 50, very poor; 100, poor; 200, fair; 300, good; 500, very good; and >1000, excellent.65 66 The optimal T/CS-SB measure will be determined using mixed model linear regression. The test of ρ²=0 (α=0.05) for five normally distributed covariates (SB plus five adjustment covariates) will have 80% power to detect a ρ² of 0.015.
Aim 2. There are no generally accepted guidelines to estimate sample size for cluster analysis. At a minimum, 10 times the number of clustering variables has been recommended. With two observations and assuming a conservative ICC of 0.4, a sample size of 202 is required to detect a medium direct and a small indirect effect (mediation) with 80% power. The sample size increases to 341 when both the direct and indirect effects are small.
Aim 3. The mixed model multiple linear regression test of ρ²=0 (α=0.05) for 16 normally distributed covariates will have 80% power to detect a ρ² of 0.021.
Results and discussion
Expected results and discussion
Aim 1
Based on emerging literature, it is anticipated that time spent in certain CS-SB will exhibit stronger associations with CMD risk scores relative to T-SB.67 68 Specifically, we hypothesise that screen time will be most deleterious, based on the stronger associations seen between TV viewing (relative to other CS-SB) and CMD risk in general adult populations.69 70 A recent meta-analysis in young adults revealed that incidence of hypertension and dyslipidaemia significantly increased with each additional hour of cumulative TV watching.71 However, it is important to note these studies relied on retrospective self-report, which is prone to recall bias,72 highlighting the critical need for this current study which uses a combination of objective SB measurement and EMA.
Aim 2
It is expected that a portion of the CMD risk associated with CS-SB will be mediated by co-occurring lifestyle behaviours. For example, if a strong association is observed between TV viewing and CMD risk (in line with aim 1), unhealthy food and alcohol consumption, which often occurs with TV viewing, may help explain this relationship.67 73 74 Nevertheless, the overall hypothesis for aim 2 is that CS-SB will continue to significantly associate with CMD risk following adjustment for clustered lifestyle behaviours.
Aim 3
We anticipate a range of intrapersonal, interpersonal and physical environmental SEM factors will be associated with T-SB and CS-SB. Specifically, for intrapersonal factors, we hypothesise that individuals who report greater psychological stress and burnout will be more sedentary.75 In terms of interpersonal factors, we anticipate those who report poor social connectedness with family, friends and their university community will engage with higher volumes of sedentary behavior.75 Finally, for the physical environment level, we hypothesise those who are less aware of their engagement with SB and exist in dwellings that are more conducive to sedentary lifestyles will be more sedentary.76 77 In summary, CMD is a significant and worsening threat to public health in CBYA.14 18 SB is strongly associated with increased CMD risk in CBYA,73 78 and therefore there is a critical need for effective interventions to reduce SB in this population. This longitudinal study will (1) provide novel insights into the association between CS-SB and CMD, (2) examine whether the clustering of lifestyle behaviours that co-exist with CS-SB drives this association and (3) identify socioecological predictors of CS-SB. Findings will support the development of an evidence-based, multilevel intervention to target SB reduction and mitigate CMD risk in CBYA. Our results will also provide a foundation for moving beyond vague public health messaging like ‘sit less, move more’.
Ethics statements
Patient consent for publication
Acknowledgments
We thank Mark Belio, Thomas Geaney, Sophia Kaczynski, Christopher Ethan Grice, Matthew Cooper and Jacob Terry-Edmunds for their assistance in data collection and database creation.
References
Footnotes
X @justinbmoorephd
Contributors JCD wrote the original manuscript and was responsible for incorporating author edits and reviewing the final draft. JCD is also involved in project management and data management/visualisation. AJC reviewed and edited the manuscript and is presently involved in project management. EC and PPL were both involved in the reviewing and editing of the manuscript as well as project management of the study. CP and JFS were involved in the reviewing and editing of the manuscript. MLM, FCL, JBM, EDH and LS were involved in the study’s conception as well as the reviewing and editing of the manuscript. EDH is the guarantor of this project.
Funding Research reported in this publication was supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health under Award Number R01HL162805. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Sponsors were not involved with study design and will not be involved with data collection, analysis, interpretation, manuscript writing or the decision to submit manuscripts for publication.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Provenance and peer review Not commissioned; peer reviewed for ethical and funding approval prior to submission.