Associations of leisure screen time with cardiometabolic biomarkers in college-aged adults

Abstract

We examined whether screen time was associated with cardiometabolic disease (CMD) risk factors in young adults. Ninety-five adults (19.9±11.4 y) self-reported medical and health behavior history, screen time (television viewing, video games and computer games), and dietary intake. Waist circumference, blood pressure, fasting glucose and lipid levels, cardiorespiratory fitness (VO₂peak), and body composition were measured. Total sedentary behavior and moderate-to-vigorous physical activity (MVPA) were measured by accelerometer. On average, leisure screen time (2.0±1.6 h·day⁻¹) accounted for 24% of total sedentary time (8.4±1.5 h·day⁻¹). After adjustment for demographics, smoking, sleep duration, total energy intake, total sedentary time and MVPA, a 1-standard deviation (SD) increase in leisure screen time was associated with a 26% higher BMI, 29% higher waist circumference, 25% higher fat mass, 23% higher triglyceride, and 24% lower VO₂peak (p<0.05). Our findings suggest that screen time may contribute to the risk of obesity and CMD in young adults.

Introduction

The college years are a time of transition for young adults during which they establish independent lifestyle and health behaviors. This transition involves a significant life change for many students as this is often the first time living independently. In the US there is a significant increase in the prevalence of obesity as adolescents (20.6%) transition into young adults (35.7%), a time period that coincides with the college years (Hales et al. 2017). Longitudinal data in college students confirm a significant increase in overweight and obesity during the first three (Gropper et al. 2012a) and four years (Gropper et al. 2012b) in college. Concurrent with this rise in overweight and obesity is a rise in cardiometabolic disease (CMD) risk factors. Estimates in college students suggest the prevalence of having at least one CMD risk factor is high, ranging from 28 to over 50% (Arts et al. 2014; Tran et al. 2017), which markedly increases lifetime risk for cardiovascular disease (Lloyd-Jones et al. 2006). Yet few studies have examined modifiable behaviors, particularly the role of sedentary behaviors, associated with development of these risk factors in a college population.

A growing body of evidence suggests that time spent in sedentary behaviors (e.g., sitting, television viewing), a construct that is independent from time spent in physical activity, is associated with an increased risk for CMD in children and youth (Saunders et al. 2014) and middle-aged to older adults (Young et al. 2016). These data are lacking, however, in young adults. Prolonged television viewing, a type of sedentary behavior classified as screen time, has been associated with a greater risk of obesity (Hu et al. 2003), diabetes (Hu et al. 2003), cardiovascular disease events and all-cause mortality (Garcia et al. 2019). Further, prolonged television viewing in adolescence, and increases in this behavior over time, have been associated with unfavorable levels of cardiovascular disease risk factors in young adulthood (Grøntved et al. 2014). In recent decades, there has been a shift in screen time with decreases in television viewing and increases in leisure computer use and video games (Saunders et al. 2014). However, the research investigating screen time and CMD risk have largely focused on television viewing only. Notably, research suggests that screen time may be more closely associated with CMD risk in children and youth than total sedentary time because screen time may promote unhealthy behaviors such as excess caloric intake (Saunders et al. 2014). To our knowledge, no studies to date have investigated the effects of multiple types of leisure screen time on CMD risk factors in young adults, particularly during the college years. This is a question of growing importance with the rising prevalence of screen time, obesity, and CMD risk factors during this time period and potential for leisure screen time to be a useful target for interventions to reduce sedentary behaviors in college-aged adults. We hypothesized that the associations between screen time and CMD risk factors for young adults would be similar to that of children and older adults; however, understanding these associations in young adults is important for developing meaningful interventions specific to this age group, particularly with the rapidly changing types of screen time activities. The purpose of the study was to determine whether leisure screen time was associated with individual CMD risk factors, independent of covariates that may mediate these relationships, such as smoking, total energy intake, and total time spent in sedentary behavior and moderate-to-vigorous physical activity (MVPA).

Methods

Participants

A convenience sample of 95 young men and women aged 18 to 25 years were recruited to participate in this cross-sectional study. Participants were recruited from a university and its surrounding community in the Pacific Northwest via flyers, email announcements, word-of-mouth, and social media. Participants were excluded from the study if they reported any of the following: diagnosed cardiovascular, metabolic, or systemic disease; antihypertensive or lipid-lowering medication use; pregnant or breast feeding; irregular menstrual cycles; collegiate athlete; or any injury limiting physical activity. All participants completed a pre-screening questionnaire and health history questionnaire to determine eligibility to participate. This study was approved by the University’s Institutional Review Board and all participants signed an informed consent.

Sequence of tests

Participants were invited to the laboratory on two occasions for a series of tests. For the first visit, participants were instructed to abstain from food and caffeine for 2 h and exercise for 12 h prior to the visit. During this visit, testing included anthropometric measurements, blood pressure measurements, a cardiorespiratory fitness measurement (VO₂peak), and a 24 h dietary recall. At the end of the visit, the participant was instructed on the accelerometer protocol and asked to wear the accelerometer for seven consecutive days. Additionally, a 24 h dietary recall, conducted over the phone, was scheduled between visits. The second visit was scheduled in the morning after an overnight fast and after the participant wore the accelerometer for seven consecutive days. During the second visit, measurements included a fasting blood draw, body composition assessment, 24 h dietary recall, survey for domain-specific sedentary behavior, and an accelerometer adherence check.

Anthropometric and blood pressure measurements

Height was recorded to the nearest 0.1 cm as the average of two measurements using a scale-mounted stadiometer (Seca 220; Hamburg, Germany). Body mass was obtained using a calibrated digital scale to the nearest 0.1 kg (Seca 220; Hamburg, Germany). Waist circumference, measured at the level of the iliac crest using a tension-regulated tape (Alimed; Dedham, MA), was recorded as the average of two measurements within 0.5 mm. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Normal weight, overweight, and obese were defined as a BMI of 18.5 to 24.9 kg·m⁻², 25.0 to 29.9 kg·m⁻², and ≥30.0 kg·m⁻², respectively.

After five minutes of seated rest in a quiet room, two readings of blood pressure, separated by two minutes, were averaged (Omron HEM-907XL; Kyoto, Japan). If the first two readings differed by more than 5 mmHg, additional readings were taken until two readings within 5 mmHg were obtained.

Cardiorespiratory fitness measurements

VO₂peak was measured using a continuous exercise test on a treadmill (TrackMaster X425C; Full Vision, Newton, KS) and computerized metabolic system (TrueOne 2400; Parvo Medics, Salt Lake City, UT). The metabolic cart was calibrated before each measurement according to the manufacturer’s instructions and standardized for barometric pressure, temperature, and humidity. After measuring resting expired gases for two minutes, a 2-minute warm up was performed at 5.6 km·h⁻¹ with a 0% grade. The treadmill speed was then increased to a comfortable but fast jogging pace for two minutes, as determined by the participant (8.0 to 11.3 km·h⁻¹). This self-selected speed remained constant for the remainder of the test. Every minute thereafter, the treadmill grade was increased by 1% each minute until volitional fatigue. During the exercise test oxygen consumption, carbon dioxide production, ventilation, heart rate, and respiratory rate were measured. Heart rate was continuously recorded using a heart rate monitor and receiver integrated with the metabolic cart (Polar Electro, Lake Success, NY). Standard criteria of respiratory exchange ratio greater than or equal to 1.10 and failure of heart rate to increase with increases in workload were used to confirm that a maximal effort was reached (American College of Sports Medicine 2017). VO₂ data were smoothed with a 15-breath moving average (Robergs et al. 2010), and VO₂peak was recorded as the highest VO₂ obtained during the last minute of exercise.

Dietary Recall

Two weekday and one weekend day 24-h dietary recalls with assessment of supplements, within one week, were obtained using a computer-based software application that facilitates recalls in a standardized fashion (Nutrition Data System for Research 2015, University of Minnesota, Minneapolis, MN). Dietary intake was gathered using a multiple pass interview approach using five distinct passes to provide multiple opportunities for the participant to recall intake. Dietary recall interviews were completed in person at each visit and once over the phone. Total energy intake was averaged across the three days and used as a covariate in the statistical analyses.

Objective measurements of activity levels

The ActiGraph GT3X+ (ActiGraph, LLC, Pensacola, FL) was used to objectively assess total time spent in sedentary behavior and physical activity. The test–retest reliability of the ActiGraph accelerometer has been previously assessed in adults and results showed intra-class correlation coefficients of 0.70–0.90 (Sirard et al. 2011). Participants with at least 6 days (4 week days and 2 weekend days) of at least 10 h·day⁻¹ of data during waking hours were included in the analysis. Data were collected at 30 Hz, converted to 60 s epochs and processed using ActiLife 6.3.1 software (ActiGraph, LLC, Pensacola, FL). Non-wear time was defined as intervals of at least 90 minutes of consecutive zero counts with allowance of a 2-minute interval of counts greater than zero with the up/downstream 30 minute consecutive zero counts window (Choi et al. 2011).

During the first visit, participants were instructed on proper placement and use of the accelerometer and were asked to maintain normal activity patterns. Participants wore the accelerometer on the anterior axillary line of the right hip during all waking hours for seven consecutive days. Participants were also provided an activity log to record structured daily activities and times when the accelerometer was removed.

The cut-points used for analysis were from data derived and validated from this model of accelerometer, specifically in young adults. The cut-points reflect the use of information obtained from all three axes of measurement of the accelerometer (vertical, anteroposterior, medio-lateral) and body position, in the form of the vector magnitude. Sedentary time was defined as all activities <150 counts·minute⁻¹ (Kozey-Keadle et al. 2011; Peterson et al. 2015). Light-intensity physical activity was defined as 150 to 2689 counts·minute⁻¹ and MVPA was defined as ≥2690 counts·minute⁻¹ (Sasaki et al. 2011). All levels of activity were calculated as the total accumulated minutes per day within the cut-points using 60 second epochs.

Fasting blood samples

A blood draw was performed on each participant during the morning hours after an 8 to 12 h fast. A small sample of blood (40 μL) was obtained from the middle or ring finger and analyzed for fasting total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, triglycerides and glucose concentrations under strict standardized operating procedures using the Cholestech LDX System (Alere Inc., Waltham, MA, USA). Independent studies indicate that the Cholestech LDX system has excellent reproducibility with standard clinical laboratory measurement of plasma lipids and lipoproteins (Shephard et al. 2007; Dale et al. 2008) and meets the National Cholesterol Education Program Adult Treatment Panel (NCEP-ATP) III criteria for accuracy and reproducibility (Bachorik and Ross 1995). Cut-points for identifying the number of CMD risk factors were defined by the criteria set forth by the NCEP-ATP III Guidelines for diagnosing metabolic syndrome, a precursor to cardiovascular disease and included waist circumference >102 cm in men and >88 cm in women, triglycerides ≥150 mg·dL⁻¹, HDL cholesterol <40 mg·dL⁻¹ in men and <50 mg·dL⁻¹ in women, blood pressure ≥130 mmHg systolic or ≥85 mmHg diastolic, and fasting glucose ≥100 mg·dL⁻¹ (National Cholesterol Education Program ATP III 2002). Lipid accumulation product was calculated using waist circumference and triglyceride levels, as previously described (Kahn 2005), and has been shown to be a better indicator of CMD risk than body mass index (Kahn 2005, 2006; Ioachimescu et al. 2010).

Body composition assessment

Air displacement plethysmography measurements were carried out using the BOD POD^® body composition system (COSMED, Rome, Italy). Participants wore a tight-fitting swimsuit or compression shorts and standard swim cap during the measurement. The system was calibrated before each measurement both empty and with a 50-liter metal cylinder inside. For the procedure, participants were weighed on the electronic scale, seated within the 450-liter chamber, and asked to remain still with hands positioned on the thighs, breathing normally. A minimum of two consecutive body volume measurements were taken and averaged. If the two volumes differed by more than 150 ml, a third measurement was conducted. Thoracic gas volume was measured using a filter and breathing tube while the participant was seated within the chamber. Body density was calculated with thoracic gas volume considered in the calculation. The Siri (general population) or Schutte (African American) equation was used to convert body density to percent body fat.

Surveys

A self-reported health history questionnaire was used to assess medical history, family health history, behavioral health history, cigarette smoking status (current, previous, or never smoker), and average number of hours of sleep per night (sleep duration). Self-reported sedentary behavior was obtained using the Sedentary Behavior Questionnaire, which has been found to have acceptable measurement properties for use among normal weight and overweight adults (Rosenberg et al. 2010). The questionnaire assessed the amount of time spent in nine different sedentary behaviors (watching television; playing computer or video games; sitting while listening to music; sitting and talking on the phone; doing paperwork or office work; sitting and reading; playing a musical instrument; doing arts and crafts; and sitting in a car, bus, or train). The questionnaire prompts participants with the following language on the first page “On a typical weekday, how much time do you spend (from when you wake up until you go to bed) doing the following?”. The second page includes the same language but asks about weekend days. These statements are followed by the nine different sedentary behavior categories. Response options include none, 15 minutes or less, 30 minutes, 1 hour, 2 hours, 3 hours, 4 hours, 5 hours, and 6 hours or more for each of the nine categories of sedentary behavior. Time spent in each of the nine categories was converted to hours, summed and then multiplied by the number of days the activity was performed (5 for weekdays and 2 for weekend days). For total weekly sedentary behavior, weekday and weekend days were summed. The total amount of time spent in leisure screen time was calculated using the total time spent in two items from the scale – watching television and playing computer or video games.

Statistical Analysis

Characteristics of the sample were summarized with mean and standard deviation (SD) for continuous variables and frequency and percentage of the study sample for categorical variables. Screen time was examined as a continuous variable (per one standard deviation [1-SD] increment) and as a categorical variable (tertiles) to determine if the associations were stronger by level of screen time or if there was a threshold effect, which has been reported in children (Saunders and Vallance 2017). A paired t-test was used to determine the difference in self-reported total sedentary time and objectively-measured sedentary time. Kruskal-Wallis (categorical data) and analysis of variance were used to determine differences in descriptive characteristics across groups. Analysis of covariance was used to determine the difference between means of CMD risk factors by tertile of screen time, controlling for non-modifiable risk factors (age, sex, race/ethnicity, and family history of type 2 diabetes).

Multivariable linear regression, using the enter method, was used to determine the independent associations between leisure screen time, as a continuous and categorical variable, and individual CMD biomarkers. The initial model (Model 1) was adjusted for non-modifiable risk factors including age, sex, race/ethnicity and family history of type 2 diabetes. Model 2 included Model 1 plus smoking status, average sleep time, and total energy intake. Model 3 included Model 2 plus accelerometer-derived minutes per day spent in sedentary behavior, moderate-to-vigorous physical activity, and wear time. When body fat and fat free mass were the outcome variables, all models additionally included height to correct for body size. These analyses were also conducted with fat free mass or fat mass entered as covariates instead of height, and results were not materially different. Therefore, and because height explained a greater proportion of the variance, we elected to use height in the models. Regression diagnostics were performed and indicated there were no problems with multicollinearity among independent variables (e.g., tolerance, condition index, and variance inflation factor). IBM SPSS^® Statistics v24 (IBM, Armonk, NY, USA) was used for all analyses with an alpha level of 0.05 used to determine statistical significance.

Results

The study cohort characteristics are presented in Table 1. Overall, the mean age was 19.9 years, 59% of participants were women, and participants represented 34 different majors across the university. The majority (82%) of participants were non-Hispanic white, 10% were Hispanic or Latino, 6% were Asian and 2% African American. On average, participants were normal weight, with a BMI of 23.9 kg·m⁻². Based on BMI, 33 participants (35%) were classified as overweight and three (3%) obese. Forty-seven participants (44%) had at least one CMD risk factor and 53 (56%) were risk factor free. Thirty-one participants (33%) had one risk factor, 10 participants (11%) had two risk factors, and one participant had three or more risk factors. The most common risk factor was low HDL cholesterol, with 21 participants (22%) exhibiting this risk factor, followed by high waist circumference (14%), triglycerides (10.5%) and systolic blood pressure (6%). The average self-reported leisure screen time was 2.0 h·day⁻¹, with no difference (p=0.83) in self-reported total sedentary time (8.7 h·day⁻¹) compared to objectively-measured total sedentary time (8.4 h·day⁻¹).

Table 1.

Participant Characteristics (n=95)

Variable	Sedentary Behavior Tertiles
	All	1 (n=32)	2 (n=36)	3 (n=27)	P
Age (years)	19.9 (1.4)	19.7 (1.5)	19.9 (1.4)	19.9 (1.2)	0.78
Women (%)	58.9	68.8	55.6	51.9	0.37
Race/ethnicity (%)					0.87
Non-Hispanic White	82.1	84.4	80.6	81.5
Hispanic	9.5	12.5	11.1	3.7
African American	2.1	0	5.6	0
Asian	6.3	3.1	2.8	14.8
Family History of T2D (%)	32.6	34.4	41.7	18.5	0.28
Body mass (kg)	70.7 (11.9)	67.6 (10.1)	70.0 (11.1)	75.4 (13.8)	0.04
BMI (kg·m−2)	23.9 (2.9)	22.9 (2.4)	23.8 (2.6)	25.4 (3.2)	<0.01
Sleep (h·day−1)	7.1 (1.0)	7.1 (0.9)	7.1 (1.1)	7.1 (1.2)	0.95
Waist (cm)	82.1 (8.2)	79.5 (6.5)	81.3 (7.4)	86.2 (9.4)	<0.01
Systolic blood pressure (mmHg)	113 (10.8)	113 (9.6)	111 (10.1)	115 (12.7)	0.34
Diastolic blood pressure (mmHg)	63 (7.8)	63 (7.0)	63 (8.4)	64 (7.9)	0.84
VO2peak (L·min−1)	3.4 (0.9)	3.2 (0.7)	3.4 (0.9)	3.4 (1.1)	0.65
VO2peak (mL·kg−1·min−1)	47.1 (7.6)	47.7 (6.4)	48.4 (7.5)	44.8 (8.6)	0.16
Total cholesterol (mg·dL−1)	157.0 (25.7)	158.0 (24.8)	159.4 (29.2)	152.5 (21.8)	0.56
HDL cholesterol (mg·dL−1)	54.5 (15.4)	55.2 (16.9)	54.9 (16.9)	53.3 (11.6)	0.88
LDL cholesterol (mg·dL−1)	83.6 (25.0)	86.8 (31.5)	87.2 (22.2)	75.8 (20.1)	0.18
Triglycerides (mg·dL−1)	92.5 (44.1)	85.1 (34.5)	83.4 (38.0)	113.2 (55.0)	0.01
Glucose (mg·dL−1)	85.1 (8.2)	85.2 (8.2)	84.6 (8.0)	85.6 (8.6)	0.89
LAP (cm·mmol−1·L−1)	24.0 (20.1)	19.2 (11.3)	20.4 (15.7)	34.5 (28.6)	<0.01
Body fat (%)	21.6 (8.9)	21.1 (7.6)	20.9 (8.7)	22.9 (10.6)	0.64
Fat mass (kg)	15.0 (6.5)	14.0 (5.2)	14.3 (5.6)	17.0 (8.4)	0.15
Fat free mass (kg)	55.3 (12.3)	53.2 (11.1)	55.2 (11.9)	57.8 (14.0)	0.37
Dietary intake
Total energy intake (kcal·day−1)	2365 (789)	2307 (634)	2367 (878)	2427 (849)	0.85
Fat intake (grams·day−1)	94 (35)	88 (30)	93 (37)	103 (36)	0.27
Carbohydrate intake (grams·day−1)	284 (111)	283 (88)	287 (118)	280 (128)	0.97
Protein intake (grams·day−1)	100 (54)	103 (59)	97 (53)	101 (51)	0.90
Survey data
Leisure screen time (h·day−1)	2.0 (1.6)	0.4 (0.3)	1.8 (0.6)	4.0 (1.0)	<0.01
Total sedentary time (h·day−1)	8.7 (3.8)	7.4 (3.8)	8.1 (3.4)	10.9 (3.5)	<0.01
Accelerometer data
Sedentary time (h·day−1)	8.4 (1.5)	8.3 (1.6)	8.4 (1.5)	8.6 (1.3)	0.73
Light activity (h·day−1)	6.0 (1.2)	6.1 (1.3)	5.9 (1.2)	5.9 (1.0)	0.73
MVPA (h·day−1)	1.2 (0.9)	1.2 (0.4)	1.2 (0.5)	1.2 (0.4)	0.96
Wear time (h·day−1)	15.6 (1.2)	15.6 (1.2)	15.5 (1.2)	15.7 (1.1)	0.75
Adherence (days)	6.9 (0.2)	7.0 (0)	6.9 (0.3)	7.0 (0)	0.45

T2D, type 2 diabetes; BMI, body mass index; VO₂peak, peak oxygen consumption; HDL, high-density lipoprotein; LDL, low density lipoprotein; LAP, lipid accumulation product; MVPA, moderate-to-vigorous physical activity. p<0.05 represents statistical significance across tertitles. Note: SI conversion factors: To convert glucose, cholesterol, and triglycerides to mmol·L⁻¹, multiply values by 0.0555, 0.0259, and 0.0113, respectively.

Risk factors by tertile of leisure screen time

After adjusting for age, sex, race/ethnicity and family history of type 2 diabetes, mean levels of BMI, waist circumference, triglycerides, fat mass, and lipid accumulation product were significantly higher across increasing tertiles of leisure screen time (p<0.05), whereas VO₂peak was significantly lower across increasing tertiles of leisure screen time (p=0.01; Figure 1). When compared to the lowest tertile, the highest leisure screen time tertile had a 9.5% (2.2 kg·m⁻²) higher BMI, 7.9% (6.3 cm) higher waist circumference, 34% (28.9 mg·dL⁻¹) higher triglyceride level, 32% (4.3 kg) higher fat mass, 91% (16.7 cm·mmol⁻¹·L⁻¹) higher lipid accumulation product, and 8% (3.9 mL·kg·min⁻¹) lower VO₂peak (p<0.05). There were no differences in mean levels of blood pressure, glucose, or total, HDL or LDL cholesterol across tertiles of leisure screen time.

Figure 1. Adjusted mean (±SE) cardiometabolic disease risk factors by tertile of leisure screen time.

P values are for linear trends; a, significantly different from tertile 1; b, significantly different than tertile 2. VO₂peak, peak oxygen consumption; LAP, lipid accumulation product. Tertile cutpoints (h·day⁻¹): T1 < 0.8; T2 0.8 to 2.6; T3 >2.6. Means adjusted for age, sex, race/ethnicity, and family history of type 2 diabetes.

Multivariable linear associations between leisure screen time and individual CMD risk factors

Multivariable-adjusted linear regression models were used to determine the independent associations between leisure screen time and CMD risk factors (Table 2). With adjustment for age, sex, race/ethnicity, and family history of type 2 diabetes (Model 1), a 1-SD increase in leisure screen time (1.6 h) was associated with a 27% SD higher BMI (0.8 kg·m⁻²; p=0.008), 32% SD higher waist circumference (2.6 cm; p=0.002), 28% SD higher triglycerides (12.3 mg·dL⁻¹; p=0.01), 28% SD higher fat mass (1.8 kg; p=0.002), 19% SD higher body fat percentage (1.7%; p=0.009), 33% SD higher lipid accumulation product (6.6 cm·mmol⁻¹·L⁻¹; p=0.002), 26% SD lower VO₂peak (1.2 mL·kg⁻¹·min⁻¹; p<0.001) and 33% SD lower LDL cholesterol (8.2 mg·dL⁻¹; p=0.008). The associations remained significant in the fully adjusted model (Model 3, p<0.05 for all), except LDL, which was attenuated to non-significance (p=0.051). There were no significant associations between leisure screen time and blood pressure, glucose, total cholesterol, or HDL cholesterol (p>0.05).

Table 2.

Multivariable associations of leisure screen time and individual cardiometabolic disease risk factors.

Variable	Model 1			Model 2			Model 3
	St β	B (SE)	P	St β	B (SE)	P	St β	B (SE)	P
Body mass index	0.27	0.07 (0.03)	0.008	0.28	0.07 (0.03)	0.007	0.26	0.07 (0.03)	0.011
Waist circumference	0.32	0.23 (0.07)	0.002	0.32	0.24 (0.08)	0.002	0.29	0.21 (0.08)	0.006
Systolic blood pressure	0.04	0.04 (0.10)	0.691	0.05	0.05 (0.10)	0.641	0.03	0.03 (0.10)	0.737
Diastolic blood pressure	0.04	0.03 (0.08)	0.393	0.05	0.03 (0.07)	0.663	0.01	0.01 (0.07)	0.994
Peak oxygen consumption	- 0.26	−0.18 (0.05)	<0.001	−0.25	−0.18 (0.05)	<0.001	−0.24	−0.17 (0.05)	0.001
Total cholesterol	- 0.14	−0.32 (0.26)	0.223	−0.13	−0.30 (0.26)	0.25	−0.12	−0.26 (0.26)	0.323
Triglycerides	0.28	1.13 (0.44)	0.013	0.28	1.14 (0.45)	0.013	0.23	0.94 (0.45)	0.039
HDL-cholesterol	0.05	0.07 (0.14)	0.635	0.05	0.07 (0.14)	0.626	0.04	0.06 (0.14)	0.701
LDL-cholesterol	- 0.33	−0.75 (0.28)	0.008	−0.30	−0.69 (0.28)	0.017	−0.25	−0.57 (0.29)	0.051
Glucose	0.04	0.03 (0.08)	0.733	0.05	0.04 (0.05)	0.640	0.08	0.06 (0.08)	0.486
Lipid accumulation product	0.33	0.62 (0.20)	0.002	0.34	0.62 (0.20)	0.002	0.29	0.53 (0.20)	0.008
Fat mass*	0.28	0.16 (0.05)	0.002	0.28	0.17 (0.05)	0.002	0.25	0.15 (0.05)	0.005
Fat free mass*	0.03	0.04 (0.06)	0.515	0.04	0.04 (0.06)	0.504	0.04	0.05 (0.06)	0.397
Body fat percentage	0.19	0.15 (0.06)	0.009	0.19	0.15 (0.06)	0.009	0.17	0.14 (0.06)	0.022

St β, standardized coefficient; B, slope; SE, standard error; HDL, high-density lipoprotein; LDL, low-density lipoprotein

^*height added into model 1 for body size; Model 1 was adjusted for age, sex, race/ethnicity and family history of type 2 diabetes. Model 2 included Model 1 plus smoking status, average sleep time, and total energy intake. Model 3 included Model 2 plus accelerometer-derived minutes per day spent in sedentary behavior, moderate-to-vigorous physical activity, and wear time.

Multivariable linear associations of tertiles of leisure screen time and individual CMD risk factors

Multivariable-adjusted regression analyses were performed using tertiles of leisure screen time to determine if the associations were stronger by level of screen time or if there were threshold effects (Table 3). After adjustment for age, sex, race/ethnicity and family history of type 2 diabetes (Model 1), and compared to the first tertile, the third tertile of leisure screen time was associated with a significantly higher BMI (B[95%CI]: 2.6[0.92, 3.6], p=0.001), waist circumference (6.3[2.4, 10.2], p=0.002), triglycerides (28.9[5.6, 52.2], p=0.015), lipid accumulation product (16.7[6.5, 26.9], p=0.002), fat mass (4.3[1.5, 7.0], p=0.003), body fat percentage (3.7[0.56, 6.76], p=0.021), and lower VO₂peak (−3.9[−6.5, −1.3], p=0.004) and LDL cholesterol (−16.4[−31.9, −0.8], p=0.04). These associations, except for LDL cholesterol (p=0.08), were slightly attenuated but remained significant after the addition of smoking, sleep duration, and total energy intake (Model 2, p<0.05). In the fully adjusted model (Model 3), and compared to the first tertile, the third tertile of leisure screen time was associated with a significantly higher BMI (2.2[0.8, 3.6], p=0.003), waist circumference (5.6[1.6, 9.7] p=0.007), lipid accumulation product (14.7[4.3, 25.2], p=0.006), fat mass (3.9[1.0, 6.7], p=0.008), body fat percentage (3.4[0.24, 6.5], p=0.035) and lower VO₂peak (−4.1[−6.6, −1.5], p=0.002). The association with triglycerides was attenuated to non-significance (23.7[−0.4, 47.8], p=0.05) with the addition of total sedentary behavior and MVPA in Model 3.

Table 3.

Multivariable linear associations of tertile of leisure screen time with individual cardiometabolic disease risk factors

	Model 1		Model 2		Model 3
	St β	P	St β	P	St β	P
Body mass index
Tertile 2	0.086	0.427	0.090	0.416	0.095	0.388
Tertile 3	0.349	0.002	0.347	0.002	0.341	0.003
Waist circumference
Tertile 2	0.074	0.505	0.077	0.499	0.085	0.450
Tertile 3	0.332	0.004	0.331	0.005	0.313	0.007
Systolic blood pressure
Tertile 2	−0.114	0.302	−0.111	0.320	−0.121	0.286
Tertile 3	0.023	0.840	0.021	0.853	0.008	0.942
Diastolic blood pressure
Tertile 2	0.014	0.902	−0.015	0.902	−0.013	0.908
Tertile 3	0.036	0.762	0.048	0.686	0.012	0.916
Peak oxygen consumption
Tertile 2	−0.042	0.603	−0.020	0.805	−0.031	0.692
Tertile 3	−0.243	0.004	−0.252	0.002	−0.242	0.002
Total cholesterol
Tertile 2	0.023	0.857	−0.006	0.857	−0.006	0.963
Tertile 3	−0.090	0.478	−0.078	0.478	−0.067	0.601
Triglycerides
Tertile 2	−0.011	0.926	−0.009	0.941	−0.007	0.952
Tertile 3	0.280	0.026	0.286	0.024	0.242	0.054
HDL-cholesterol
Tertile 2	0.024	0.827	0.028	0.800	0.030	0.791
Tertile 3	0.060	0.586	0.057	0.614	0.053	0.639
LDL-cholesterol
Tertile 2	−0.051	0.715	−0.067	0.636	−0.063	0.653
Tertile 3	−0.299	0.040	−0.265	0.085	−0.224	0.132
Glucose
Tertile 2	−0.071	0.573	−0.070	0.573	−0.059	0.625
Tertile 3	−0.040	0.752	−0.042	0.752	−0.025	0.835
LAP
Tertile 2	0.050	0.668	0.051	0.670	0.063	0.591
Tertile 3	0.362	0.003	0.361	0.004	0.329	0.006
Fat mass
Tertile 2	0.081	0.404	0.068	0.493	0.071	0.477
Tertile 3	0.288	0.004	0.293	0.004	0.271	0.008
Fat free mass
Tertile 2	0.005	0.929	0.016	0.782	0.019	0.742
Tertile 3	0.065	0.269	0.061	0.298	0.072	0.206
Body fat percentage
Tertile 2	0.067	0.399	0.052	0.516	0.052	0.515
Tertile 3	0.188	0.021	0.193	0.019	0.172	0.035

St β, standardized beta; Referent category, tertile 1; Tertile cutpoints (h·day⁻¹): T1 < 0.8; T2 0.8 to 2.6; T3 >2.6. HDL, high-density lipoprotein; LDL, low-density lipoprotein; LAP, lipid accumulation product. Model 1 was adjusted for age, sex, race/ethnicity and family history of type 2 diabetes. Model 2 included Model 1 plus smoking status, average sleep time, and total energy intake. Model 3 included Model 2 plus accelerometer-derived minutes per day spent in sedentary behavior, moderate-to-vigorous physical activity, and wear time.

Discussion

This cross-sectional study evaluated the associations of leisure screen time with individual CMD risk factors in a sample of college students. We found that leisure screen time was positively associated with BMI, waist circumference, triglycerides, fat mass and lipid accumulation product and negatively associated with VO₂peak. Notably, these associations were independent of relevant covariates including smoking status, sleep duration, total energy intake, total sedentary behavior and moderate-to-vigorous physical activity. These findings suggest that leisure screen time may contribute to the risk of obesity and CMD in young adults.

We found robust associations between leisure screen time and several markers of obesity, which is consistent with the literature. Shibata et al. (Shibata et al. 2016) reported increases in television viewing over 12 years were associated with greater waist circumference in Australian adults age 25 years and older. Hu et al. (Hu et al. 2003) reported sedentary behaviors, particularly television viewing, were associated with a significantly elevated risk of obesity in over 50,000 women from the Nurses’ Health Study. Several systematic reviews (Rhodes et al. 2012; Biddle et al. 2017) on sedentary behavior concluded that the association between sedentary behavior and markers of obesity is complex and the evidence suggests stronger, more consistent associations with total screen time than other types of sedentary behavior, when predicting obesity. It may be that domain-specific sedentary behavior is associated with obesity and that total sedentary behavior or television viewing alone may not be predictive of risk. We extend previous findings to young adults and include other screen time behaviors that are common in young adults including computer and video games. A novel aspect of our study is that we show screen time is positively associated with multiple indicators of obesity including BMI, fat mass, waist circumference and lipid accumulation product, a measure of total body lipid over-accumulation. Together these data suggest that leisure screen time may be an important target for obesity prevention efforts in college-aged adults.

Notably, our data show that leisure screen time was negatively associated with VO₂peak, which is the gold standard measurement of cardiorespiratory fitness. Indeed, compared to participants in the first tertile of leisure screen time, on average, participants in the third tertile had an 8% lower VO₂peak. VO₂peak is widely accepted as a strong, independent risk factor for CMD and premature mortality and small increases in fitness can significantly reduce risk of cardiovascular morbidity and mortality (Blair et al. 1989; Mandsager et al. 2018). For example, large population-based studies show that low levels of cardiorespiratory fitness are associated with a high burden of CMD risk factors in adults (Aspenes et al. 2011; Erez et al. 2015). Similarly, data from the CARDIA study showed baseline cardiorespiratory fitness in young adults was inversely associated with the risk of developing hypertension, diabetes, and metabolic syndrome in middle age (Carnethon et al. 2003). Finally, data from the Aerobics Center Longitudinal Study indicate that low fitness accounts for more deaths than other risk factors including obesity, smoking, high cholesterol, and diabetes (Blair 2009). To date, few studies have investigated screen time and fitness, with most studies in youth and reporting mixed findings (Bai et al. 2016; Porter et al. 2017; Motte et al. 2019; Cabanas-Sénchez et al. 2019). Given the importance of VO₂peak as a predictor of CMD morbidity and mortality, future studies should confirm our findings in larger, more diverse samples.

Our findings are consistent with previous studies showing associations between television viewing time and CMD risk factors (Thorp et al. 2010; Pinto Pereira et al. 2012; Grøntved et al. 2014; Saunders et al. 2014; Biddle et al. 2017). Our data indicate that as leisure screen time increases from the first to third tertile, biomarkers of CMD increase correspondingly. In regression analysis, when compared to the first tertile, the highest tertile of leisure screen time was associated with significantly poorer metabolic biomarkers, suggesting there may be a threshold amount of screen time that is associated with significantly higher CMD risk in college-aged adults. Our data suggest a threshold level of >2.6 h·day⁻¹ adversely affects CMD biomarkers, including triglycerides, VO₂peak, and multiple markers of obesity. Our data are in line with current screen time guidelines for children that recommend no more than 2 h·day⁻¹ of screen time and provide preliminary evidence of a threshold of screen time that increases prevalence of unfavorable CMD risk factors in young adults.

We found the associations between leisure screen time and CMD risk factors were independent of total sedentary behavior and moderate-to-vigorous physical activity. Adjustment for these variables only slightly attenuated the results, suggesting that mechanisms other than those induced by moderate-to-vigorous physical activity are contributing to the association between screen time and CMD biomarkers. Leisure screen time has been associated with other unhealthy behaviors such as excess food intake. Although we included total energy intake into the models, we did not assess dietary intake while performing leisure screen time behaviors or macronutrient content. Longitudinal studies are needed to assess the types of leisure screen time and accompanying behaviors that may contribute to an increased risk in young adults.

Average leisure screen time in our participants was 2 h·day⁻¹, which is about 1 h·day⁻¹ lower than that reported in young adults in the US. National prevalence estimates of television viewing and playing computer games in young adults, aged 20 to 24 y, were 2.0 and 1.0 h·day⁻¹, respectively (Bureau of Labor Statistics 2018). Approximately 48% of young adults in the US enroll in college (Bureau of Labor Statistics 2018), which may partially explain the difference in prevalence of screen time between our study and national averages. The questions we used to assess screen time asked about time spent watching television (including videos) and playing computer or video games, but did not specifically ask about time spent texting or engaging in social media on cell phones or personal devices. As technology advances, questions to assess screen time should be updated to include various devices and activities.

Measurement of various domains of sedentary behavior via questionnaire, as well as objectively measuring total sedentary behavior with an accelerometer were strengths of the study. With advances in technology, there has been a shift in screen time in recent decades with decreases in television viewing and increases in watching videos on computers and playing computer and video games, which were captured by the survey questions used in our study. However, using personal devices for texting and social media were likely not captured and are a limitation of the study. Other limitations include a cross-sectional design, which limits our ability to establish causal inferences between screen time and CMD risk factors, scheduled dietary recalls which may have altered the dietary habits of participants, and a lack of ethnic and racial diversity of our sample. Our participants were mainly non-Hispanic white and physically active, accumulating an average of 1.2 h·day⁻¹ of moderate-to-vigorous physical activity, which may not be generalizable to all college-student or young adult populations that are not enrolled in college.

In summary, we found that leisure screen time was positively associated with BMI, waist circumference, triglycerides, fat mass, and lipid accumulation product and negatively associated with VO₂peak. These findings suggest that leisure screen time may contribute to the risk of obesity and CMD in young adults and provides preliminary evidence to support leisure screen time as a specific target for interventions to reduce sedentary behaviors in college-aged adults.