Beyond weight: associations between 24-hour movement behaviors, cardiometabolic and cognitive health in adolescents with and without obesity

ABSTRACT

Background

Adolescence is a critical time for establishing behaviors. 24-hour movement behaviors, including physical activity, sleep, and sedentary time, are likely to influence obesity, cardiovascular, and cognitive health. The aim was to examine associations between 24-hr movement behaviors, cardiometabolic health and cognitive functions in adolescents with and without obesity.

Methods

This was a cross-sectional study that included adolescents (n = 30, ages 12–16) with obesity and normal weight controls matched on age and sex. 24-hr movement behaviors of physical activity, sedentary time, and sleep were assessed using waist-worn accelerometers. Cardiometabolic health was measured using flow-mediated dilation (FMD) in the brachial artery, body composition via dual x-ray absorptiometry, blood pressure, and blood analyses of cholesterol, glucose, and insulin. Cognitive health was assessed using two computer-based tasks. Linear regressions were used to examine associations between 24-hr movement behaviors, cardiometabolic health, and cognition.

Results

In examining relationships between 24-hr movement behaviors and cardiometabolic health, when adjusted for body fat percentage, MVPA was positively associated with cardiovascular health (FMD log difference 0.1, 95%CI: 0.003, 95%CI: .001, .01, p = .020), sedentary time was negatively associated (−0.7, 95%CI: −1.3, −0.2, p = .016), and total sleep time was negatively associated with HDL cholesterol (−0.1, 95%CI: −0.2, −0.005, p = .039). There were no statistically significant associations between 24-hr movement behaviors and cognitive outcomes, except sleep and reactive control. When examining relationships between cardiometabolic and cognitive outcomes, higher HDL was associated with improved cognitive accuracy and higher insulin was associated with slower reaction times.

Conclusions

24-hour movement behaviors of MVPA, sedentary time, and sleep time were associated with cardiometabolic measurements in a small sample. 24-hr movement behaviors, particularly MVPA and sedentary time, may be important behaviors for cardiometabolic health in adolescents, independent of body composition. Additional research is needed on the triadic relationship between 24-hr movement behaviors, cardiometabolic health, and cognitive performance.

KEYWORDS:

• Cognition
• physical activity
• sleep
• sedentary

Introduction

Identifying children and adolescents at risk for overweight and obesity is a critical first step towards prevention and treatment of future health and cardiometabolic risk. However, discussion of weight and obesity with children and parents is often received with anxiety (Nnyanzi 2016) or lack of engagement. (Davidson et al. 2018) The behaviors that contribute to obesity, such as physical activity, sedentary time, and sleep, may be more accepted by parents in conversations with their primary care physician, but these behaviors also have important health consequences, independent of obesity (Steele et al. 2012). Physical inactivity has been associated with negative physical and mental health outcomes in adolescents (Hallal et al. 2006). There is increasing research that not only moderate-to-vigorous physical activity (MVPA) is important for these health outcomes but other movement behaviors within a 24-hour day, including sedentary time and sleep, have substantial impact on health. There is evidence from self-reported data that meeting 24-hr movement behavior guidelines has been associated with better cardiometabolic and cognitive health in children and adolescents (Katzmarzyk and Staiano 2017; Walsh et al. 2018). Additionally, research suggests that cardiometabolic health is likely associated with cognitive performance (Donohoe and Benton 1999; Novak and Hajjar 2010). Thus, it is likely that 24-hr movement behaviors, cognitive health, and cardiometabolic health are related as in Figure 1. Adolescence is a critical time for establishing behaviors into adulthood. Some chronic health problems, such as Type 2 diabetes (Lawrence et al. 2021) and abnormal lipid levels (Centers for Disease Control and Prevention [CDC] 2010), are increasingly being diagnosed in adolescence, supporting the important role of establishing positive movement behaviors during adolescence.

Figure 1. Hypothesized relationships between behaviors, psychological health and physical health outcomes.

Research on the effects of 24-hr movement behaviors in children on cardiometabolic health suggests that physical activity has the most consistent relationship with cardiometabolic outcomes (Katzmarzyk and Staiano 2017), but research examining all three movement behaviors among adolescents is still limited. When examined alone, physical inactivity has been associated with multiple markers of cardiometabolic risk (Pahkala et al. 2008), and recent studies including both physical activity and sedentary time have found that MVPA may be most important for physical health outcomes (Skrede et al. 2019). Improved sleep has also been associated with positive cardiometabolic outcomes in adolescence, including fasting insulin and triglycerides (Hilmisson et al. 2019).

In addition to cardiovascular health, cognitive function may also be negatively affected by physical inactivity, sedentary time, and insufficient sleep (Donnelly et al. 2016; Medic et al. 2017). A study by Walsh et al. found that meeting each additional 24-hr movement behavior guideline was associated with higher global cognition (Walsh et al. 2018). The relationship between movement behaviors and cognitive functions is likely bi-directional. Movement behaviors may influence cognitive performance through physiological mechanisms, such as brain-derived neurotrophic factor, neuroelectric activity, or blood flow (Hillman and Biggan 2017), but cognitive skills such as planning, flexibility, and problem solving are needed to maintain physical activity and sleep behaviors (Allan et al. 2016; Nelson et al. 2019). Research examining these relationships between 24-hr movement behaviors and these cognitive outcomes is limited in adolescents.

While 24-hr movement behaviors are likely related to cardiometabolic and cognitive health, there are also relationships between cardiometabolic and cognitive health. Cardiometabolic risk has been associated with poorer cognitive outcomes, but research has been focused on older adults (Femminella et al. 2018). In adolescents, Yeh et al. found that blood pressure, body-mass-index, and waist circumference were negatively associated with various academic test scores (Yeh et al. 2015). It is possible that these impaired metabolic states, including increased blood glucose and blood pressure, may impair cognitive function (Donohoe and Benton 1999; Novak and Hajjar 2010). As previously mentioned, these relationships are also likely bi-directional, as executive functions, including cognitive control, have been implicated in the ability to perform positive health behaviors and thus cardiometabolic risk (Allan et al. 2016; Nelson et al. 2019). However, no studies have examined this triadic relationship between movement behaviors, physical health, and cognitive health in adolescence.

Adolescents are increasingly being diagnosed with obesity, diabetes or pre-diabetes, dyslipidemia, and hypertension (Cook et al. 2003), and early research has shown that cardiovascular disease, including atherosclerosis, starts in childhood and adolescence (Berenson et al. 1998). Thus, novel measures of cardiovascular function may be better able to detect early signs of cardiovascular disease in adolescence, rather than waiting until adulthood to detect symptoms such as angina or decreased exercise tolerance that manifest later in the disease process. One of these measures of cardiovascular function is brachial flow-mediated dilation (FMD), which may be more sensitive than other common clinical markers, such as obesity to early change in arterial health and ultimately cardiovascular risk (Bond et al. 2015). Limited studies, however, have examined associations between FMD with any of the 24-hr movement behaviors including physical activity and sedentary time (Hopkins et al. 2012; McManus et al. 2015). Less evidence exists on the relationship between regular sleep duration and FMD, but children with suspected sleep-disordered breathing have been found to have altered vascular responses (Kontos et al. 2015).

Several studies have examined the relationship between body composition and FMD measures of cardiovascular risk in children and adolescents (Yilmazer et al. 2010; Ryder et al. 2016). Interestingly, neither of these studies included physical activity as a potential covariate despite known associations between physical activity and arterial health (Dias et al. 2015). Additionally, the majority of studies examining FMD in children and adolescents have been conducted with clinical patients, particularly those with obesity or impaired metabolism such as Type 1 diabetes (Chiesa et al. 2019). Identifying and improving these cardiovascular risk parameters may be even more important for populations who have not yet been diagnosed with obesity, while behavioral habits and more severe health outcomes are preventable or reversible.

As no studies to date have examined 24-hr movement behaviors with FMD measures of cardiometabolic risk, a study of 30 adolescents, including adolescents with and without obesity, was conducted to examine preliminary cross-sectional associations between 24-hr movement behaviors, cardiometabolic health, and cognitive health. It was hypothesized that higher physical activity, decreased sedentary time, and increased sleep would be associated with positive cardiometabolic and cognitive outcomes. Additionally, it was hypothesized that positive cardiometabolic health would be associated with better cognitive outcomes.

Materials and methods

Study design

This is a cross-sectional study including adolescents with obesity and normal weight controls matched on age and sex. Primary analyses are descriptive, with preliminary associations explored.

Participants

Adolescents (12–16 years of age) with and without obesity to ensure variation in body composition were recruited from both specific primary care referrals from a pediatric nurse practitioner specializing in obesity treatment and general recruitment from the community. Inclusion criteria were English-speaking males and females aged 12–16 years, matched for sex and age. Obesity was defined as body-mass-index higher than the 95^th percentile on standard growth charts; normal weight was considered <85^th percentile. Overweight participants with BMI-percentiles between 85^th and 95^th were not included in the current study. While puberty status was assessed using self-reported Tanner Stages, puberty was not used as an inclusion criteria as vascular function has been more strongly associated with age than pubertal status (Marlatt et al. 2013). Teens were excluded if their obesity was due to identified genetic factors, were diagnosed with metabolic or endocrine disease (i.e. diabetes), were undergoing treatment for severe psychiatric disorders, or were currently taking medications that could influence cardiovascular or endothelial function (e.g. BP medication, depression medication, and glucose metabolism). Parents or guardians provided written, informed consent, and participants provided written assent. All procedures were approved by the University of Arkansas Institutional Review Board.

Measures

24-hour movement behaviors

Movement behaviors were measured using Actigraph GT9× accelerometers on an elastic belt worn on the waist. Adolescents were asked to wear the devices for 7 days for 24 hr per day, except for water-based activities, and asked to record bed and wake times and times they removed the device. No active water activities were recorded and thus were not imputed. Bed and wake times were compared to visual inspection of activity data in Actilife software (v6.13.3) according to published heuristics (McVeigh et al. 2016) by 3 independent research assistants. Disagreements of more than 15 min were re-evaluated for a consensus. Data were then processed using Troiano non-wear algorithms (Troiano et al. 2008), Evenson cutpoints to determine intensity (Evenson et al. 2008), and the Sadeh sleep algorithm for sleep metrics (Sadeh et al. 1994). To be included as a valid day for physical activity and sedentary time, a minimum of 8-hr wear was needed and at least 3 days of valid wear were needed to be included in the study (Rich et al. 2013). The primary movement outcomes were minutes spent in moderate-to-vigorous activity, percentage of wake time in sedentary time to account for variations in total wear time and total sleep time.

Cardiometabolic Measures. Both traditional (blood pressure, cholesterol, glucose and insulin, and body fat percentage) and non-traditional (FMD) measures of cardiometabolic risk were assessed.

FMD and blood pressure

Flow-mediated dilation (FMD), a non-invasive procedure using ultrasound equipment, provides a sensitive measure of arterial compliance, and thus cardiovascular health. FMD was assessed in the brachial artery approximately 10 cm proximal to the antecubital space using standardized procedures (Restaino et al. 2016). Briefly, a validated, open-source software (FloWave.us) for beat-to-beat analysis of vessel wall detection and quantification of shear rate (Coolbaugh et al. 2016) was utilized and processed using FMD Studio (version 3.6; Quipu srl, Pisa, Italy). The FMD was calculated as (maximum diameter −baseline diameter)/baseline diameter · 100, where the maximum diameter represents the maximum diameter after 5-min distal ischemia. To account for the allometric scaling dependent on baseline diameter, the difference between the log of the maximum diameter and the log of the baseline diameter (LogDiff) was adjusted for the log of the baseline diameter in regression models (Atkinson et al. 2013).

Blood

Fasting blood analyses were collected using a private laboratory company in accordance with standard pediatric practice for adolescents with abnormal results. These included fasting insulin (Marson et al. 2016), glucose, cholesterol (Srinivasan et al. 2006), and triglycerides (Muscogiuri et al. 2017) as markers of cardiometabolic risk.

Body composition. As the primary measure of body composition that is regularly used in clinical practice, height and weight were measured by research technicians according to standardized procedures and used to calculate body mass index (BMI) and age and sex-specific BMI percentiles (Kuczmarski et al. 2000). BMI percentiles were used to classify obesity status. As a more valid measure of body composition, dual x-ray absorptiometry (Prodigy, GE Healthcare, Chicago, IL) was used to measure total body fat percentage.

Cognitive function

Inhibition was measured using a Go/No-go task (Lamm et al. 2014), and cognitive control was assessed using the AX-CPT task (Lamm et al. 2013). Both tasks were presented in a quiet room on a 17.3-in Dell laptop using E-Prime 3 software (Psychological Software Tools, Pittsburgh, PA). Accuracy and reaction time were the primary outcome variables.

Go/NoGo task

Connie Lamm et al. (2018) The current task was a modified version of the traditional letter Go/NoGo task (Conners and Staff 2000). Prior to playing the actual game, all participants completed a 20-trial practice block. The total number of trials presented was 340. All “NoGo” trials presented the letter “X.” “Go” trials consisted of all other letters except the letter “K.” Accuracy for Go and NoGo trials and reaction time on correct trials are presented.

Cognitive control task

The AX-CPT task assesses two types of cognitive control: reactive control, the ability to change action strategies based on last minute information and proactive control, the ability to actively maintain information in the face of distraction (Braver et al. 2009). The task was administered according to previous procedures (Lamm et al. 2013). Outcome variables include reaction time on correct trials, accuracy (AY accuracy indicating more reactive control and BX accuracy indicating more proactive control), and BSI error. Based on the works of Braver and colleagues (Braver et al. 2009) (supporting information) a Behavioral Shift Index (BSI) was calculated (AY-BX)/(AY+BX) for correct probe reaction times and combined cue/probe error rates to show control style (i.e. more proactive or reactive control style). More positive values indicate a proactive style of responding, while less positive values indicate a more reactive style of responding.

Procedures

Participants were asked to complete an online survey that included questions on demographics, education, as well as two items on self-reported physical activity prior to their visit to the Exercise science Research Center at the University of Arkansas. Participants were asked to arrive fast and complete informed consent/assent, body composition measures, and FMD. They were then given an ad-libitum breakfast meal prior to completing cognitive testing. Participants were then fitted with an accelerometer and given instructions to give a blood sample at the private health-care facility within the next 2 weeks.

Statistical analysis

First, descriptive statistics were the primary focus of this current study. Comparisons of movement behaviors, cognitive variables, and cardiometabolic variables between non-obese and obese participants (as determined by BMI percentiles) were made using appropriate chi-square tests, t-tests, or Wilcoxon Rank Sum tests. Second, to describe the relationships between variables in the current sample (Greenland and Chow 2019), linear regressions were used to examine the associations between movement behaviors as the independent variable with cardiometabolic outcomes and cognitive outcomes as the dependent variables adjusted for age and sex. Similar regressions with cardiometabolic measures as the independent variables and cognitive performance as the dependent variables adjusted for age and sex were run. Errors were checked for normality. For variables with evidence of non-normality robust standard errors (go accuracy) or log-transformed p-values are presented with non-transformed estimates (insulin) for interpretation. To examine independent contributions of movement behaviors, regressions were additionally adjusted for body fat percentage from DXA. All available data were used for each variable. Partial eta-squared effect sizes are presented to inform future studies using the categories of 0.01 indicating a small effect, 0.06 indicating a medium effect and 0.14 indicating a large effect. All analyses were made using Stata I/C v 14.2, and the statistical significance was set at p < .05. No adjustments were made for multiple comparisons.

Results

Sample descriptives

All 30 participants completed the Go/NoGo cognitive task; however, due to technical errors, only 20 participants (9 without obesity, 11 with obesity) had completed valid data on the AX-CPT task. All 30 participants completed DXA and resting blood pressure measures. After the visit to the laboratory, 25 participants had complete blood analyses (12 without obesity, 13 with obesity; 5 participants did not attend the clinic for a follow-up blood draw). FMD measures were attempted in 30 participants but included from 15 (8 without obesity, 7 with obesity; 1 participant requested not to complete due to being uncomfortable, and images were unclear due to fidgeting/technician errors in 14 participants). Using BMI percentile groupings, 17 participants were included as non-obese and 13 participants were included as obese. However, when examining body fat percentage from DXA, five of the non-obese participants were classified as obese (Taylor et al. 2003). Twenty-nine participants had valid accelerometer data (one participant lost the accelerometer).

A summary of demographics and means/medians of each variable can be seen in Table 1. Only one male participant was pre-pubertal (Tanner stage 1). The only statistically significant differences between the non-obese and obese groups were in total accelerometer wear time, body fat percentage, triglycerides, cholesterol, LDL, and maximum shear rate from FMD, with the obese group having shorter wear time and poorer cardiometabolic outcomes.

Table 1. Comparison in demographics and behaviors between obese and non-obese groups determined by BMI percentiles.

	Total	Non-obese (n=17)	Obese (n=13)	p-value
Age (years)	14.8 (1.8)	14.5 (1.6)	15.2 (1.9)	.307
Sex (% girl)	14 (46.7%)	9 (52.9%)	5 (38.5%)	.431
Race (% white) n=27	22 (81.5%)	11 (68.8%)	11 (100%)	.377
Grade n=29				.287
6^th-8^th	12 (41.4%)	9 (52.9%)	3 (27.3%)
9^th-10^th	9 (31.0%)	5 (29.4%)	4 (36.4%)
11^th-12^th	8 (27.6%)	3 (17.6%)	5 (45.5%)
Parent Ed (% college degree)	18 (62.1%)	12 (70.1%)	6 (54.6%)	.387
Movement Behaviors
Accelerometer (n=29) Days wear	6.9 (1.3)	7.3 (0.7)	6.5 (1.7)	.139
Weartime (min/day)	849.1 (72.7)	872.8 (62.1)	819.9 (76.3)	.049
Vector magnitude (counts per min)	611.3 (181.1)	630.8 (208.1)	587.3 (145.8)	.531
Sedentary Time (%)	69.0% (7.6%)	69.4% (8.4)	68.6% (6.8)	.773
Moderate Physical Activity (min/day)	13.1 (5.5, 17.9)	13.8 (7.4, 20.8)	9.0 (4.7, 14.0)	.188
Vigorous Physical Activity (min/day)	0 (0, 2.3)	0.6 (0, 6.6)	0 (0, 1.3)	.422
MVPA (min/day)	13.3 (5.5, 20.7)	14.8 (9.3, 30.0)	10.1 (5.3, 15.7)	.148
Total Sleep Time (min/day)	488.5 (51.8)	480.8 (54.4)	498.0 (48.9)	.385
Sleep Efficiency (%)	95.5 (2.6)	96.0 (2.4)	94.9 (2.7)	.289
Self-report PA (n=27) % meeting PA recs	8 (29.6%)	6 (37.5%)	2 (18.2%)	.302
% no sports nor PE	7 (25.9%)	4 (25%)	3 (27.3%)
Cognitive Health
Go/No Go Task (n=30) Go accuracy %	97.7 (3.0)	98.1 (2.6)	97.2 (3.4)	.460
No go accuracy %	64.7 (16.8)	66.9 (16.0)	61.7 (18.1)	.409
Go/No Go Correct RT	329.8 (307.8, 372.7)	326.1 (309.7, 364.3)	347.2 (304.1, 379.9)	.851
Go/No Go Incorrect RT	277.6 (266.5, 292.9)	277.1 (269.4, 291.1)	281.8 (264.6, 305.5)	.851
AX-CPT (n=20) BSI Error	0.5 (0.3)	0.5 (0.2)	0.5 (0.4)	.872
BSI Correct RT	0.3 (0.2)	0.4 (0.1)	0.3 (0.2)	.216
BX Accuracy %	82.5 (9.8)	82.3 (10.1)	82.8 (10.0)	.178
AY Accuracy %	41.3 (26.7)	48.6 (23.9)	32.2 (28.5)	.912
Cardiometabolic Health
BMI percentile	78.5 (48, 94)	50 (43, 61)	96 (91, 97)	<.001
BMI	23.8 (6.0)	19.9 (2.2)	20.0 (5.5)	<.001
% Body fat (n=30)	30.4 (11.2)	24.1 (8.9)	38.7 (7.9)	<.001
Insulin (μIU/mL) (n=25)	7.0 (5.6, 11.9)	6.7 (5, 7.6)	11.6 (6.2, 19.2)	.888
Glucose (mg/dL) (n=25)	88.2 (7.0)	86.6 (6.8)	90.2 (7.0)	.205
Triglycerides (mg/dL) (n=25)	87.5 (44.5)	66.9 (24.6)	113.6 (51.3)	.006
Cholesterol (mg/dL) (n=25)	153.2 (30.3)	141.1 (25.9)	168.5 (29.4)	.021
LDL(mg/dL) (n=25)	87.5 (25.6)	78.4 (23.2)	99.1 (24.7)	.042
HDL(mg/dL) (n=25)	47.9 (8.5)	48.7 (7.9)	46.8 (9.5)	.591
Resting SBP (mmHg) (n=30)	110.6 (8.9)	108.0 (7.0)	114.1 (10.1)	.061
Resting DBP (mmHg) (n=30)	65.6 (6.2)	65.5 (7.1)	65.6 (5.1)	.971
FMD% (n=15)	7.7 (5.3, 12.2)	6.2 (5.3, 12.5)	9.5 (5.2, 11.7)	.896
Max shear rate per second (n=15)	533.8 (198.3)	473.1 (186.7)	700.6 (128.9)	.044

N (%), chi; Mean (SD), ttest; Median (25^th-75^th), Wilcoxon.

MVPA, moderate-to-vigorous physical activity; RT, reaction time; FMD%, Flow-mediated dilation calculated as (maximum diameter −baseline diameter)/baseline diameter · 100; BSI, Behavioral Shift Index with more positive values indicating a proactive style of responding, while less positive values indicating more reactive style of responding; AY Accuracy, more reactive control; BX Accuracy, more proactive control; LogDiff, the difference between the log of the maximum diameter and the log of the baseline diameter.

Associations between movement behaviors and cardiometabolic outcomes

MVPA was negatively associated with insulin (−0.2, 95%CI: −0.4, −0.01, p = .024), triglycerides (−1.1, 95%CI: −2.1, −0.1, p = .028) and positively associated with %FMD (0.4, 95%CI: 0.1, 0.7, p = .011) and FMD Log difference (.003, 95%CI: .001, .006, p = .015) as seen in Table 2. Sedentary time was negatively associated with FMD log difference (−0.7, 95%CI: −1.2, −0.2, p = .011). Sleep time was positively associated with glucose (0.06, 95%CI: .002, 0.1, p = .043) and negatively associated with HDL (−0.09, 95%CI: −0.2, −0.02, p = .016). There were no associations between 24-hr movement behaviors and body fat percentage. Estimated effect sizes can be seen in Table 4. When significant models were additionally adjusted for body fat percentage, the associations between MVPA and insulin (−0.1, 95%CI: −0.3, 0.05, p = .079), triglycerides (−0.8, 95%CI: −1.7, 0.2, p = .104), and FMD (0.1, 95%CI: −0.3, 0.5, p = .599 were attenuated as well as the relationship between total sleep time and glucose (0.04, 95%CI: −0.03, 0.1, p = .209). However, the relationship between MVPA and FMD Log difference (0.1, 95%CI: 0.003, 95%CI: .001, .01, p = .020), sedentary time and FMD Log difference (−0.7, 95%CI: −1.3, −0.2, p = .016), and total sleep time and HDL (−0.1, 95%CI: −0.2, −0.005, p = .039) remained statistically significant when additionally adjusted for body fat percentage.

Table 2. Linear Regressions between behaviors (independent) and cardiometabolic and cognitive outcomes (dependent) adjusted for sex and age.

		MVPA (min/day)		Sedentary (%)		Sleep (min/night)
		B (95%CI)	p-value	B (95%CI)	p-value	B (95%CI)	p-value
Cardiometabolic	% Body fat (n=29)	−0.2 (−0.5, 0.03)	.081	3.7 (−60.1, 67.5)	.906	0.8 (−0.01, 0.2)	.079
	Insulin (μIU/mL) (n=24)^c	-0.2 (−0.4, −.01)	.024^a	.44 (−0.1, 1.0)	.056^a	0.07 (−0.02, 0.2)	.890^a
	Glucose (mg/dL) (n=24)	−0.1 (−0.3, .05)	.168	0.2 (−0.2, 0.6)	.373	0.06 (.002, 0.1)	.043
	Triglycerides (mg/dL) (n=24)	-1.1 (−2.1, −0.1)	.028	1.3 (−1.7, 4.2)	.383	0.4 (−0.02, 0.9)	.059
	Cholesterol (mg/dL) (n=24)	−0.6 (−1.3, 0.2)	.145	0.2 (−2.0, 2.3)	.880	0.2 (−0.2, 0.5)	.376
	LDL (mg/dL) (n=24)	−0.5 (−1.1, 0.2)	.135	0.3 (−1.6, 2.1)	.765	0.2 (−0.1, 0.4)	.283
	HDL (mg/dL) (n=24)	0.1 (−0.05, 0.3)	.142	−0.3 (−0.8, 0.1)	.155	-0.09 (−0.2, −0.02)	.016
	Resting SBP (mmHg) (n=29)	0.1 (−0.3, 0.4)	.746	−0.1 (−0.6, 0.3)	.502	0.05 (−0.01, 0.1)	.105
	Resting DBP (mmHg) (n=29)	−0.0003 (−0.2, 0.2)	.997	0.1 (−0.2, 0.4)	.562	0.01 (−0.04, 0.06)	.581
	FMD% (n=15)	0.4 (0.1, 0.7)	.011	−0.6 (−1.4, 0.2)	.142	−.007 (−0.1, 0.1)	.889
	LogDiff^c (n=15)	.003 (.001, .006)	.015	-0.7 (−1.2, −0.2)	.011	−.0002 (−.001, .001)	.726
	Max shear per second (n=15)	1.2 (−9.0, 11.1)	.798	−374.9 (−2207.9, 1457.1)	.661	0.03 (−3.0, 3.1)	.981
Cognitive	Go Accuracy (%)^c	−0.04 (−0.1, 0.03)	.250	9.7 (−3.3, 22.6)	.137	−0.01 (−0.04, 0.01)	.311
	No go Accuracy (%)	−0.2 (−0.5, 0.1)	.240	20.5 (−52.4, 93.5)	.567	−0.05 (−0.2, 0.05)	.297
	Reaction Time (correct Go)	−0.8 (−2.1, 0.5)	.206^a	33.2 (−286.8, 353.3)	.873^a	−.003 (−0.5, 0.5)	.988^a
	BSI Error (n=19)	.0004 (−.007, .008)	.907	−0.3 (−2.4, 1.7)	.728	.002 (−.002, .005)	.310
	BSI Correct RT (n=19)	.003 (−.002, .008)	.183	−0.2 (−1.6, 1.3)	.812	−.0006–.003, .002)	.550
	AY Accuracy (n=19)^c	.003 (−0.001, .001)	.102	0.3 (−1.5, 2.1)	.745	-.002 (−.003, −.0003)	.023
	BX Accuracy (n=19)^c	.001 (−0.001, .003)	.384	−0.01 (−0.7, 0.6)	.975	.0002 (−.001, .001)	.805

^ap-values calculated from comparison of log-transformed data.

^badjusted for log baseline.

^crobust standard errors.

MVPA, moderate-to-vigorous physical activity; RT, reaction time; FMD%, Flow-mediated dilation calculated as (maximum diameter −baseline diameter)/baseline diameter · 100; BSI, Behavioral Shift Index with more positive values indicating a proactive style of responding while less positive values indicating more reactive style of responding; AY Accuracy, more reactive control; BX Accuracy, more proactive control; LogDiff, the difference between the log of the maximum diameter and the log of the baseline diameter.

Table 3. Linear Regressions between cardiometabolic (independent) and cognitive (dependent) outcomes adjusted for sex and age.

	Go Accuracy (%)^a		No Go Accuracy (%)		Reaction time (ms)
	B (95%CI)	p-value	B (95%CI)	p-value	B (95%CI)	p-value
% Body fat (n=30)	−0.07 (−0.2, 0.03)	.157	−0.4 (−0.8, 0.1)	.124	1.3 (−0.7, 3.2)	.196
Insulin (μIU/mL) (n=25)	−0.06 (−0.2, 0.06)	.327	0.05 (−0.7, 0.8)	.895	3.2 (0.3, 6.1)	.033
Glucose (mg/dL) (n=25)	−0.04 (−0.03, 0.2)	.722	−0.2 (−1.3, 0.8)	.626	1.6 (−2.8, 6.1)	.452
Triglycerides (mg/dL) (n=25)	−0.02 (−0.05, 0.01)	.154	−0.4 (−0.2, 0.1)	.515	0.2 (−0.4, 0.9)	.406
Cholesterol (mg/dL) (n=25)	−0.03 (−0.06, 0.003)	.076	0.02 (−0.2, 0.2)	.866	0.5 (−0.4, 1.3)	.248
LDL (mg/dL) (n=25)	−0.02 (−0.06, 0.01)	.182	−0.01 (−0.2, 0.2)	.907	0.4 (−0.6, 1.4)	.430
HDL (mg/dL) (n=25)	−0.07 (−0.2, 0.08)	.335	0.8 (0.01, 1.5)	.047	2.1 (−1.4, 5.6)	.223
Resting SBP (mmHg) (n=30)	−0.01 (−0.1, 0.1)	.843	−0.1 (−0.8, 0.6)	.789	−0.3 (−3.4, 2.7)	.839
Resting DBP (mmHg) (n=30)	−0.07 (−0.2, 0.07)	.334	0.2 (−0.6, 1.1)	.564	0.7 (−2.8, 4.2)	.684
FMD% (n=15)	0.04 (−0.2, 0.2)	.660	−0.6 (−1.9, 0.8)	.371	−2.1 (−7.0, 2.7)	.352
LogDiff b (n=15)	5.1 (−18.8, 29.1)	.644	−42.3 (−174.3, 89.8)	.492	−155.2 (−579.3, 269.0)	.434
Max shear rate per second (n=15)	−.002 (−0.01, 0.01)	.568	.003 (−0.05, 0.5)	.888	.05 (−0.1, 0.2)	.556

^arobust errors.

^badjusted for log baseline.

MVPA, moderate-to-vigorous physical activity; RT, reaction time; FMD%, Flow-mediated dilation calculated as (maximum diameter −baseline diameter)/baseline diameter · 100; BSI, Behavioral Shift Index with more positive values indicating a proactive style of responding while less positive values indicating more reactive style of responding; AY Accuracy, more reactive control; BX Accuracy, more proactive control; LogDiff, the difference between the log of the maximum diameter and the log of the baseline diameter

Table 4. Estimate effect sizes for relationships between 24-hr movement behaviors, cardiometabolic, and cognitive variables.

		MVPA (min/day)	Sedentary (%)	Sleep (min/night)	Go Accuracy (%)	No Go Accuracy (%)	Go Correct Reaction time (ms)
Cardiometabolic	% Body fat (n=29)	0.19	0.00	0.12	0.08	0.09	0.06
	Insulin (μIU/mL) (n=24)^a	0.19	0.13	0.12	0.04	0.00	0.20
	Glucose (mg/dL) (n=24)	0.09	0.04	0.19	0.01	0.01	0.03
	Triglycerides (mg/dL) (n=24)	0.22	0.04	0.17	0.13	0.02	0.03
	Cholesterol (mg/dL) (n=24)	0.10	0.00	0.04	0.12	0.00	0.06
	LDL (mg/dL) (n=24)	0.11	0.00	0.06	0.06	0.00	0.03
	HDL (mg/dL) (n=24)	0.44	0.10	0.26	0.04	0.17	0.07
	Resting SBP (mmHg) (n=29)	0.00	0.02	0.10	0.00	0.00	0.00
	Resting DBP (mmHg) (n=29)	0.04	0.01	0.01	0.02	0.01	0.01
	FMD% (n=15)	0.52	0.31	0.00	0.01	0.07	0.08
	LogDiff^b (n=15)	0.46	0.49	0.01	0.02	0.05	0.06
	Max shear per second (n=15)	0.01	0.02	0.00	0.01	0.00	0.03
Cognitive	Go Accuracy (%)^c	0.05	0.06	0.05
	No go Accuracy (%)	0.05	0.00	0.11
	Reaction Time (correct Go)	0.06	0.00	0.03
	BSI Error (n=19)	0.00	0.00	0.26
	BSI Correct RT (n=19)	0.12	0.03	0.02
	AY Accuracy (n=19) ^c	0.08	0.01	0.14
	BX Accuracy (n=19) ^c	0.02	0.00	0.01

Effect size is partial eta-squared

MVPA, moderate-to-vigorous physical activity; RT, reaction time; FMD%, Flow-mediated dilation calculated as (maximum diameter −baseline diameter)/baseline diameter · 100; BSI, Behavioral Shift Index with more positive values indicating a proactive style of responding, while less positive values indicating more reactive style of responding; AY Accuracy, more reactive control; BX Accuracy, more proactive control; LogDiff, the difference between the log of the maximum diameter and the log of the baseline diameter

Associations between movement behaviors and cognitive outcomes

There were no statistically significant associations between 24-hr movement behaviors and cognitive outcomes (Table 2), except a negative association between AY accuracy and total sleep time (−0.0002 min/night, 95%CI: −0.003, −0.0003, p = .023).

Associations between cognitive outcomes and cardiometabolic outcomes

HDL was positively associated with accuracy during the NoGo trial during the Go/NoGo task (0.8, 95%CI: 0.01, 1.5), p = .047) as seen in Table 3. Insulin was positively associated with reaction time during the correct Go trials during the Go trial (3.2, 95%CI: 0.3, 6.1, p = .033).

Discussion

This study examined the triadic relationship between 24-hr movement behaviors, cardiometabolic health, and cognitive performance among adolescents. The current study found associations between 24-hr movement behaviors and cardiometabolic outcomes in adolescents. More physical activity and less sedentary time were associated with better cardiometabolic function, independent of weight-based obesity status. There were minimal associations between 24-hr movement behaviors and cognitive tasks. Among the relationships between cardiometabolic and cognitive outcomes, higher HDL was associated with better accuracy on the cognitive task. This study provides preliminary data on these associations; however, conclusions are limited due to missing data and small sample sizes. These are novel findings among adolescents that highlight relationships between device-based measured 24-hr movement behaviors, cognitive health, and cardiometabolic outcomes that can potentially be used to expand health assessments beyond weight measures.

The relationship between higher MVPA and lower sedentary time with positive cardiometabolic outcomes was expected (Katzmarzyk and Staiano 2017), however the majority of these relationships were attenuated when adjusted for percent body fat. Some of these associations between 24-hr movement behaviors and cardiometabolic health, however, remained even when adjusted for body fat percentage, suggesting the effects of movement behaviors on cardiometabolic health are independent of obesity. In particular, higher MVPA and lower sedentary time were associated with improved arterial health measures of flow-mediated dilation. This is one of the first studies to examine these associations of early cardiovascular risk with 24-hr movement behaviors. Even in the small sample, large effect sizes were observed for physical activity with body fat percentage, insulin, HDL, and FMD measures; sedentary time and FMD measures, sleep time and glucose; no go accuracy and HDL. As there is still limited research to establish the meaningfulness of changes in FMD, the clinical meaningfulness of these effects is currently unknown and longer-term studies are needed to confirm health benefits. Physical activity may be a more proximal health behavior to beneficial cardiometabolic outcomes; physical inactivity typically precedes the development of obesity. While increased adiposity may have a stronger relationship with blood markers of impaired metabolism (e.g. insulin resistance), arterial health may be impacted earlier (i.e. before insulin resistance occurs). Thus, FMD measures, even in non-obese adolescents, may be able to detect risk earlier compared to traditional clinical markers in the cardiometabolic risk cascade. Substantial training or designated technicians may be needed, however, for staff to establish reliable skill in FMD measures.

Practically, these behaviors may be important for improving vascular health and cardiometabolic risk in adolescents regardless of obesity status. Both aerobic and resistance activities have been shown to improve vascular function in children and adolescents. A review of exercise interventions with FMD outcome measures in children or adolescents with obesity found six studies that employed either aerobic only or a combination of aerobic and resistance training (Dias et al. 2015). While there was wide diversity in the exercise protocols, overall exercise interventions improved FMD and vascular function. A more recent review found nine studies examining the effects of exercise on cardiometabolic outcomes including FMD in children and adolescents with obesity and found inconclusive evidence on the effect of aerobic or aerobic combined with resistance exercise on FMD (Salamt et al. 2019). When considering what types of physical activity may be most beneficial for improving FMD, both moderate and high intensity acute physical activity has recently been shown to impact arterial function (McManus et al. 2019). A single study found resistance training over 10 weeks improved vascular function as measured by FMD in non-obese adolescents (Yu et al. 2016). More research is needed on the ideal type of exercise, in addition to sedentary time and sleep interventions, to improve FMD in adolescents. Importantly, any intervention to engage families in physical activity or obesity programs should be tailored to the individual family and include adolescents in the decision-making process (Banks et al. 2014).

Contrary to hypotheses, minimal significant associations were found between 24-hr movement behaviors and cognitive outcomes, however, there were limited relationships between cardiometabolic variables and cognitive outcomes, including an effect that suggests a more reactive style of responding is associated with less sleep time. As cognitive performance is a complex outcome, a larger sample may be needed to control for other factors influencing cognitive performance. However, these preliminary findings suggest that cardiometabolic health may influence cognitive performance even in adolescents.

The current study adds to the literature using FMD measures in adolescents by including both device-based 24-hr movement behaviors and cognitive performance measures. Both obese and non-obese participants were included for a range of bodyweight and cardiometabolic profiles. However, these findings need to be repeated in a larger sample as several participants had incomplete measures. These non-compliance issues have been evident in previous studies of adolescents in a non-clinical community sample (Howie and Straker 2016). Future studies may trial field-based measures of cardiometabolic health. However, field measurements of body composition may compromise validity. As a result of the small sample, adjustments for multiple comparisons and robust fitting of regression models to include additional confounders were not made. With a larger sample size, compositional data analysis approaches may be used to understand the related activity behaviours within a 24-hr time period, as physical activity, sedentary time, and sleep are inherently non-independent. (Pedišić 2014) In the current study, 5 participants who were classified as normal weight based on BMI z-score cutoffs were classified as obese from DXA results (Taylor et al. 2003). More detailed measures of body composition, such as body fat percentage, may be needed to better understand the effects of body composition on health. More sensitive body fat percentage measures can be used to inform adolescents and parents who may assume they are not at increased risk of health problems due to normal BMI status but are at additional risk from increased adiposity.

Conclusion

This study found evidence that among a small sample of adolescents with and without obesity, 24-hr movement behaviors of physical activity, sedentary time, and sleep were associated with early measures of arterial health, independent of adiposity. Additionally, poorer cardiometabolic health may be associated with poorer cognitive performance. Improving obesity-related behaviors, including physical activity, sedentary time, and sleep, may improve both cardiometabolic and cognitive health, however larger studies are needed to confirm these findings. Health-care professionals and practitioners should encourage physical activity and sleep hygiene among adolescents, regardless of obesity status. However, in the reverse direction, poor cognitive control may negatively impact these behaviors and cardiometabolic health. Strategies to improve both 24-hr movement behaviors and cardiometabolic health in adolescents are needed to prevent future increasing rates of cardiometabolic diseases and declining cognitive health in populations in adulthood.

Implications for practice

Even adolescents who are not obese, as measured by BMI or DXA but are physically inactive, may be at risk for impaired cardiovascular function. Thus, health-care practitioners should assess 24-hr movement behaviors, including physical activity, in adolescents regardless of weight status, and potentially examine additional cardiometabolic risk factors if the patient is inactive. US National Physical Activity Guidelines recommend adolescents get a minimum of 60 min per day of physical activity, and teens who do not meet these guidelines should be counseled to increase their physical activity.

Acknowledgments

We sincerely thank all participants and parents for their participation and Matthew Paxton for assisting in data collection. This work was funded by an ABI Biomedical Research Grant from the Arkansas Biosciences Institute and supported by an Honors College Research Grant.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The work was supported by the Arkansas Biosciences Institute .