An Examination of the Effect of Sleep Quality and TV/Computer

Usage on Stress: Based on National Health and Nutrition

Examination Survey Data

Xingzhi Xu

School of Statistics, University of Minnesota-Twin Cities, Minneapolis, Minnesota, 55414, U.S.A.

Keywords: Stress, TV Usage, Computer Usage, Sleep Quality, NHANES, Interaction Effect.

Abstract: This study examines the effect of TV/computer usage and sleep quality on stress on a large population with

comprehensive demographic characteristics. It also aimed to identify the difference of treatment effect

between the female group and the male group. This study includes 5092 participants in the National Health

Nutrition Examination Survey (NHANES). These participants were asked to answer questionnaires regarding

TV/computer usage, sleep quality, and mental health. We applied a binary logistic regression and a Poisson

regression. TV usage, computer usage, and sleep quality were positively associated with both the odds of

experiencing stress-related symptom and the frequency of stress-related symptoms. No difference of treatment

effect was found between the female group and the male group for sleep quality. The female group was found

to be less affected by TV usage. Effect of computer usage was detected on the male group, but not on the

female group. A reduction in TV/computer usage and an improvement of sleep quality can be performed to

reduce the frequency and the odds of experiencing stress-related symptoms. Future study with comprehensive

population groups and standardized measurement for sleep quality is needed to reduce bias.

1 INTRODUCTION

Anxiety is an emotion characterized by feelings of

tension, worried thoughts and physical changes like

increased blood pressure (Kazdin 2000). Study has

also shown that anxiety can be spread in social

networking sites (Seabrook 2016). Many of us have

experience stress and anxiety during COVID-19

pandemic (Ahmed 2020). It is a serious matter since

anxiety is shown to be related to the risk of

cardiovascular disease (Tully 2016). There has been

research shown that TV/computer viewing has

positive correspondence to anxiety-related problems

(Maras 2015, Werneck 2021, Yu 2019). On the other

hands, sleep quality is also highly associated to

anxiety (Al-Khani 2019, Ghrouz 2019). However,

they either targeted specific population group, or

considered a single factor that leads to anxiety. In this

research, we want to examine the effect of both non-

regular sleeping and time of computer/TV use on

mental health, and the population in this study have

comprehensive demographic characteristics. We will

be using binary logistic regression and Poisson

regression; and the factors we considered are TV

usage, computer usage, how often overly sleepy

during day, ever reported trouble sleeping, and sleep

time (details listed in Table 1). We aim to determine

the effect of those factors on the occurrence and

frequency of stress-related symptoms, and we hope

that with the large sample size, this study can give us

a more accurate result compares to previous studies.

2 METHODS

2.1 Study Design and Participants

The data used in this study came from the National

Health and Nutrition Examination (Centers for

Disease Control and Prevention 2015-2016). The

National Health and Nutrition Examination Survey

(NHANES) is a program of studies designed to assess

the health and nutritional status of adults and children

in the United States. The survey is unique in that it

combines interviews and physical examinations.

NHANES is a major program of the National Center

for Health Statistics (NCHS). NCHS is part of the

Centers for Disease Control and Prevention (CDC)

410

Xu, X.

An Examination of the Effect of Sleep Quality and TV/Computer Usage on Stress: Based on National Health and Nutrition Examination Survey Data.

DOI: 10.5220/0011371400003438

In Proceedings of the 1st International Conference on Health Big Data and Intelligent Healthcare (ICHIH 2022), pages 410-416

ISBN: 978-989-758-596-8

and has the responsibility for producing vital and

health statistics for the Nation.

The NHANES program began in the early 1960s

and has been conducted as a series of surveys

focusing on different population groups or health

topics. In 1999, the survey became a continuous

program that has a changing focus on a variety of

health and nutrition measurements to meet emerging

needs. The survey examines a nationally

representative sample of about 5,000 persons each

year. These persons are located in counties across the

country, 15 of which are visited each year.

The NHANES interview includes demographic,

socioeconomic, dietary, and health-related questions.

The examination component consists of medical,

dental, and physiological measurements, as well as

laboratory tests administered by highly trained

medical personnel.

Findings from this survey will be used to

determine the prevalence of major diseases and risk

factors for diseases. Information will be used to assess

nutritional status and its association with health

promotion and disease prevention. NHANES

findings are also the basis for national standards for

such measurements as height, weight, and blood

pressure. Data from this survey will be used in

epidemiological studies and health sciences research,

which help develop sound public health policy, direct

and design health programs and services, and expand

the health knowledge for the Nation. The present

study includes participants who were 18-year-old and

older at the baseline measurement (N = 5735). A total

of 643 participants who had missing data (599) or

answered, “I don’t know/choose not to say” (44) are

excluded (N = 5092). The demographic characteristic

of the 5092 participants is listed in Table 1.

2.2 Predictors

The predictors included in this study is listed in the

last 5 rows in Table 1. The first 3 predictors (TV

usage, computer usage, how often overly sleepy

during day) are set to be continuous variables. The

last 2 predictors (ever reported trouble sleeping, sleep

time) are set to be binary categorical variables.

Interactions are also introduced to capture the

interplay between each variable.

2.3 The Scale of Stress Level (Response)

In the original data from NHANES, the codebook for

Mental Health – Depression Scanner has 10

categories: 1. “Having little interest in doing things”,

2. “Feeling down, depressed, or hopeless”, 3.

“Trouble sleeping or sleeping too much”, 4. “Feeling

tired or having little energy”, 5. “Poor appetite or

overeating”, 6. “Feeling bad about yourself”, 7.

“Trouble concentrating on things”, 8. “Moving or

speaking slowly or too fast”, 9. “Thought you would

be better off dead”, 10. “Difficulty these problems

have caused”. Each category has 4 levels: 0. Not at

all, 1. Several days, 2. More than half the days, 3.

Nearly every day.

In order to transform the codebook for mental

health into a response variable, we sum up the levels

of each category (except the 10th category, because

of too many missing values) and create a response

ranging from 0 to 27, details in row 5, Table 1. To

reduce outliers, we collapse all response with 10 or

more into 10.

A binary response is set to be whether participants

have experienced any stress-related symptoms, which

is either 0 or 1. A count response is set to be the scale

of stress level (i.e., the frequency of stress-related

symptoms), which ranges from 0 to 10.

2.4 Statistical Analysis

The Statistics Analysis Tool R (R Core Team 2020)

and the package glmnet (Friedman 2010) were used

for data analysis. Binary logistic regression analysis

was performed for the association between the

predictors (computer and TV times, whether overly

sleepy, ever reported trouble sleeping, normal or

abnormal sleep time) and the binary response

(whether having stressed or not), which determines

whether a participant is having stress-related

symptoms or not. Poisson regression was also

performed for the association between the predictors

and stress level for those participants with stress,

whichdetermine how stress a participant is for those

who are determined to be having stress-related

symptoms. A Fixed significance level of p ≤ 0.05 was

used for the analysis. The package xtable (Dahl 2019)

was used to produce coefficient table.

An Examination of the Effect of Sleep Quality and TV/Computer Usage on Stress: Based on National Health and Nutrition Examination

Survey Data

411

Table 1: Demographic characteristics of the study population.

CHARACTERISTICS

MEAN ± SD /

FREQUENCY (%)

AGE (YEARS) 48.2 ± 18.5

GENDER Male 2486 (48.8%)

Female 2606 (51.2%)

RACE Mexican American 924 (18.1%)

Other Hispanic 652 (12.8%)

Non-Hispanic White 1700 (33.4%)

Non-Hispanic Black 1081 (21.2%)

Non-Hispanic Asian 540 (10.6%)

Other Race and Multi-Racial 195 (3.8%)

EDUCATION LEVEL Less than 9th grade 535 (10.5%)

9-12th grade 659 (12.9%)

High school graduate/GED or equivalent 1176 (23.1%)

College or AA degree 1512 (27.0%)

College graduate or above 1209 (23.7%)

Don't know 1 (0.01%)

STRESS LEVEL (FROM 0 TO 10) None (0) 1562 (30.7%)

Slightly Stressed (1-5) 2521 (49.5%)

Stressed (6-9) 602 (11.8%)

Highly Stressed (>=10) 407 (8.0%)

TV USAGE (HOUR) Light use (0-1) 778 (27.9%)

Average use (2-4) 2922 (45.9%)

Heavy use (5+) 1392 (10.8%)

COMPUTER USAGE (HOUR) Light use (0-1) 2756 (54.1%)

Average use (2-4) 1786 (35.1%)

Heavy use (5+) 550 (10.8%)

HOW OFTEN OVERLY SLEEPY

DURING DAY

Never (0 times) 904 (17.8%)

Rarely (1 time a month) 1176 (23.1%)

Sometimes (2-4 times a month) 1677 (32.9%)

Often (5-15 times a month) 914 (17.9%)

Almost always (16-30 times a month) 421 (8.3%)

EVER REPORTED TROUBLE

SLEEPING

Yes 1383 (27.2%)

No 3709 (72.8%)

SLEEP TIME (HOUR)

Normal (7-10)

3754 (73.7%)

Abnormal (<7 or >10) 1338 (26.2%)

3 RESULTS

3.1 Association between Predictors and

Binary Response of Stress

A detailed summary of the binary logistic regression

is listed in Table 2. Notice that all predictors except

the interactions are positively associated with the

binary response of stress. For each unit increase in TV

usage, the odds of having stress symptoms is

increased by 10.71% (95% CI, 4.61%−17.2%). The

usage of PC also leads to an increase in the odds ratio

by 14.35% (95% CI, 4.96%−24.73%). Feeling overly

sleepy during day leads to a more significant increase

in the odds ratio of 73.19% (95% CI,

62.07%−85.26%). The odds ratio of sleep time is

increased by 17.33% (95% CI, -2.38%−41.25%),

ICHIH 2022 - International Conference on Health Big Data and Intelligent Healthcare

412

which means that having an abnormal sleep time (less

than 7 hours or greater than 10 hours) would result in

an increasing chance of experiencing anxiety-related

symptoms. For those who reported trouble sleeping,

they have an increase of 213.77% (95% CI,

149.71%−297.5%) to the odds ratio than those who

did not. Furthermore, the interaction between TV

usage and PC usage leads to a slight decrease in the

odds ratio of 2.46% (95% CI, 0.36%−4.52%). A more

significant decrease is related to the interaction

between sleep time and report trouble sleeping, which

leads to 37.39% (95% CI, 4.7%−58.57%).

Table 2: Binary Regression Analysis: coefficients and etc..

EXP(Β) LOWER CI UPPER CI P-VALUE

INTERCEPT 0.4747 0.3762 0.5982 0.0000

TV USAGE (TV) 1.1071 1.0461 1.1720 0.0004

PC USAGE (PC) 1.1435 1.0496 1.2473 0.0023

OVERLY SLEEPY (OS) 1.7319 1.6207 1.8526 0.0000

SLEEP TIME (ST) 1.1733 0.9762 1.4125 0.0899

REPORT TROUBLE

SLEEPING (RTS)

3.1377 2.4971 3.9750 0.0000

TV * PC 0.9754 0.9548 0.9964 0.0223

ST * RT 0.6261 0.4143 0.9530 0.0274

3.2 Association between Predictors and

Scale of Stress

A detailed summary of the Poisson regression is listed

in Table 3. Notice that similar to the binary logistic

regression, all predictors except the interactions are

positively associated with the binary response of

stress. For each hour increase in TV usage, there is a

6.47% (95% CI, 4.93%−8.04%) increase on the scale

of stress. The usage of PC also leads to a slight

increase in the scale of stress by 2.28% (95% CI,

0.02%−4.58%). Feeling overly sleepy during day

leads to a more significant increase on the scale by

31.05% (95% CI, 28.97%−33.16%). The increase on

scale of stress of sleep time is increased by 12.38%

(95% CI, 6.51%−18.53%), which means that having

an abnormal sleep time would result in an increasing

chance of experiencing anxiety-related symptoms.

For those who reported trouble sleeping, they have an

increase of 75.03% (95% CI, 67.27%−83.15%) on the

scale than those who did not.

Furthermore, both interaction terms included in

the Poisson model have slightly or insignificant

decrease on the scale of stress. The interaction

between TV usage and PC usage leads to a slight

decrease on the scale by 0.82% (95% CI,

0.3%−1.34%). A slight decrease is related to the

interaction between sleep time and report trouble

sleeping, which leads to 6.57% (95% CI, -

1.14%−13.69%).

Table 3: Poisson Regression Analysis: coefficients and etc.

EXP(Β) LOWER CI UPPER CI P-VALUE

INTERCEPT 1.0959 1.0250 1.1712 0.0071

TV USAGE (TV) 1.0647 1.0493 1.0804 0.0000

PC USAGE (PC) 1.0228 1.0002 1.0458 0.0469

OVERLY SLEEPY (OS) 1.3105 1.2897 1.3316 0.0000

SLEEP TIME (ST) 1.1238 1.0651 1.1853 0.0000

REPORT TROUBLE

SLEEPING (RTS)

1.7503 1.6727 1.8315 0.0000

TV * PC 0.9918 0.9866 0.9970 0.0022

ST * RT 0.9343 0.8631 1.0114 0.0930

An Examination of the Effect of Sleep Quality and TV/Computer Usage on Stress: Based on National Health and Nutrition Examination

Survey Data

413

3.3 Sub-group Analysis

A sub-group analysis stratified by gender is

performed and the result is listed in Table 4.

Interestingly, the female group seems to be less

affected by the usage of computer. Both binary

logistic and Poisson regressions shows insignificancy

(p-value is relatively large). The female group is also

less affected by the usage of TV compares to the male

group in the binary regression (female’s 4.72%

increase in odds ratio compares to male’s 11.7%

Table 4: Subgroup Analysis (Stratified by Gender): Exp(β)

MALE BINARY MALE POISSON FEMALE BINARY

FEMALE

POISSON

INTERCEPT 0.3731 0.9967 0.6816 1.2740

TV USAGE (TV) 1.1170 1.0508 1.0472 1.0512

PC USAGE (PC) 1.2573 1.0461 1.0108 0.9989

OVERLY SLEEPY (OS) 1.7005 1.3070 1.7427 1.2899

SLEEP TIME (ST) 1.2898 1.2054 1.2723 1.1597

REPORT TROUBLE

SLEEPING (RTS)

2.9304 1.8637 3.1398 1.7268

TV * PC 0.9562 0.9889 1.0097 0.9982

ST * RT 0.6609 0.8570 0.6715 0.9498

4 DISCUSSION

Positive association has been found for all the

predictors in both binary logistic regression and

Poisson regression, though there has been a slight

negative association for the interaction terms.

Comparing to another study (Ghrouz 2019), the

increase in both the odds ratio and on the scale of

stress for sleep time is lower (39% increase from

Ghrouz compares to 12.38% increase in this study).

This difference can be explained by the difference

definition of sleep quality between Ghrouz and this

study. In Ghrouz, the use of Pittsburgh Sleep Quality

Index (PSQI) allows the sleep quality predictor to

include different aspects such as sleep quality, sleep

latency, sleep duration, disturbance, etc. However, in

this study, the factor sleep time is only one aspect of

sleep quality, and it is more generalized. Having an

average of 7-10 hours of sleep does not guarantee of

having a high-quality sleep, and thus some

participants with normal sleep time would still

experience stress-related symptoms. The more

influential factors are report trouble sleeping and

overly sleepy during day. Participants who report

trouble sleeping to doctor has a 213.77% increase in

odds ratio and 75.03% increase on scale of stress than

those who did not. Report trouble sleeping can be

interpreted as another factor of sleep quality. Only

those who had severe trouble on sleeping would

consider seeking medical help, hence they are more

likely to have stress-related symptoms. While on the

other hand, overly sleepy during day is a sign of poor

quality of sleeping in the night. This factor is similar

to the sleep quality factor in Ghrouz’s study, and the

result of 31.05% increase in the chance of having

anxiety in this study is relatively similar to Ghrouz’s

39%. The effect of TV usage in this study is 10.71%

increase in odds ratio and 6.47% increase on the scale

of stress for each unit increase. Comparing to the

effect of TV usage in another study (Yu 2019), this

study’s result is relatively lower. It may be

contributed by the different characteristics of the

study population, that Yu’s study targeted

participants in China. It is also possible that the

difference in TV programs led to a difference

between the two studies. Yet, comparing to a

Brazilian study (Werneck 2020), their result of 30.0%

increase in anxiety by TV (4 or more hours usage) and

38.5% increase by PC (4 or more hours usage) is

closer to the present study. It is possible that

individuals who choose to do unhealthy movement

(watching TV or using computer) has the same effect

on mental health compare to individuals who are

forced to do unhealthy movement due to quarantine.

A further study is needed to confirm this hypothesis.

The sub-group analysis stratified by gender is

performed and listed in Table 4. Both the male group

and the female group seems to be equally affected by

sleep quality, and no significant difference was found

in the three factors of sleep quality. Yet, the usage of

ICHIH 2022 - International Conference on Health Big Data and Intelligent Healthcare

414

computer is found to have insignificant effect in the

female group, while the usage of TV has less effect

on female (4.72%) than male (11.7%). This finding

resembles to another Werneck’s study (Werneck

2021), where boys have higher odds ratio (1.24) than

girls’ odds ratio (1.09). It is possible that male is more

common to use computer to play games, but female

is not; hence, the purpose of using computer for male

is more likely to be entertainment, but the purpose for

female is more likely to be browsing internet, doing

work, or doing online shopping. A further study is

needed to test the effect of using computer on mental

health stratified by the purpose of using computer.

Strengths of the present study include containing

a variety of participants with different race, education

level, and gender. This study also shows results

resemble to previous studies with smaller sample

size, where male’s mental health is more affected by

the usage of TV than female.

Limitations of the present study include not using

the Pittsburgh Sleep Quality Index (PSQI) to produce

a standardized measurement for sleep quality, hence

the comparison of the effect of sleep quality to other

studies may not be accurate. Furthermore, the lack of

including the last category (“Difficulty these

problems have caused”) on the depression scanner

questionnaire may introduce biasness to the result.

Lastly, an exclusion to 643 missing or refused

responses in the dataset contributed to a 11.21%

reduction in the sample size, and hence the result may

be even more biased.

5 CONCLUSIONS

TV/computer usage and sleep quality are found to be

positively correlated to the probability of

experiencing stress-related symptoms. These factors

are also positively correlated to the frequency of

stress-related symptoms. Hence, a reduction in

TV/computer usage and an improvement of sleep

quality can be performed to reduce the frequency and

the odds of experiencing stress-related symptoms.

This study provides a more accurate view of the

effect of sleep quality and TV/computer usage on

stress by containing a large and diverse population. It

confirms results from previous studies and detect a

phenomenon, where men are more susceptible to

stress-related symptoms as the usage of TV or

computer increases. However, since the data from

NHANES does not incorporate PSQI to measure

sleep quality, this study lacks the use of standardized

measurement compares to other studies. Hence, a

future study with comprehensive population groups

and standardized measurement for sleep quality is

needed to reduce bias. Moreover, another future study

is needed to investigate why TV and computer usages

have more impact on the stressfulness of males than

females.

REFERENCES

Ahmed, N. J., Alrawili, A. S., & Alkhawaja, F. Z. (2020).

The Anxiety and Stress of the Public during the Spread

of Novel Coronavirus (COVID-19). Journal of

Pharmaceutical Research International, 32(7), 54-59.

https://doi.org/10.9734/jpri/2020/v32i730460

Al-Khani AM, Sarhandi MI, Zaghloul MS, Ewid M, Saquib

N. A cross-sectional survey on sleep quality, mental

health, and academic performance among medical

students in Saudi Arabia. BMC Res Notes. 2019 Oct

21;12(1):665. doi: 10.1186/s13104-019-4713-2. PMID:

31639038; PMCID: PMC6802108.

Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR,

Kupfer DJ. The Pittsburgh Sleep Quality Index: a new

instrument for psychiatric practice and research.

Psychiatry Res. 1989 May;28(2):193-213. doi:

10.1016/0165-1781(89)90047-4. PMID: 2748771.

Centers for Disease Control and Prevention (CDC).

National Center for Health Statistics (NCHS). National

Health and Nutrition Examination Survey Data.

Hyattsville, MD: U.S. Department of Health and

Human Services, Centers for Disease Control and

Prevention, [year 2015-2016][web URL:

https://wwwn.cdc.gov/nchs/nhanes/continuousnhanes/

default.aspx?BeginYear=2015].

David B. Dahl, David Scott, Charles Roosen, Arni

Magnusson and Jonathan Swinton (2019). xtable:

Export Tables to LaTeX or HTML. R package version

1.8-4. https://CRAN.R-project.org/package=xtable

Ghrouz AK, Noohu MM, Dilshad Manzar M, Warren

Spence D, BaHammam AS, Pandi-Perumal SR.

Physical activity and sleep quality in relation to mental

health among college students. Sleep Breath. 2019

Jun;23(2):627-634. doi: 10.1007/s11325-019-01780-z.

Epub 2019 Jan 26. PMID: 30685851.

Gould CE, Spira AP, Liou-Johnson V, Cassidy-Eagle E,

Kawai M, Mashal N, O'Hara R, Beaudreau SA.

Association of Anxiety Symptom Clusters with Sleep

Quality and Daytime Sleepiness. J Gerontol B Psychol

Sci Soc Sci. 2018 Mar 2;73(3):413-420. doi:

10.1093/geronb/gbx020. PMID: 28379498; PMCID:

PMC6074813.

Jerome Friedman, Trevor Hastie, Robert Tibshirani (2010).

Regularization Paths for Generalized Linear Models via

Coordinate Descent. Journal of Statistical Software,

33(1), 1-22. URL https://www.jstatsoft.org/v33/i01/.

Kazdin, A. E. (2000). Encyclopedia of psychology.

Washington, D.C., American Psychological

Association.

Maras D, Flament MF, Murray M, Buchholz A, Henderson

KA, Obeid N, Goldfield GS. Screen time is associated

An Examination of the Effect of Sleep Quality and TV/Computer Usage on Stress: Based on National Health and Nutrition Examination

Survey Data

415

with depression and anxiety in Canadian youth. Prev

Med. 2015 Apr; 73:133-8. doi:

10.1016/j.ypmed.2015.01.029. Epub 2015 Feb 2.

PMID: 25657166.

R Core Team (2020). R: A language and environment for

statistical computing. R Foundation for Statistical

Computing, Vienna, Austria. URL https://www.R-

project.org/.

Seabrook EM, Kern ML, Rickard NS. Social Networking

Sites, Depression, and Anxiety: A Systematic Review.

JMIR Ment Health. 2016 Nov 23;3(4): e50. doi:

10.2196/mental.5842. PMID: 27881357; PMCID:

PMC5143470.

Tully PJ, Harrison NJ, Cheung P, Cosh S. Anxiety and

Cardiovascular Disease Risk: a Review. Curr Cardiol

Rep. 2016 Dec;18(12):120. doi: 10.1007/s11886-016-

0800-3. PMID: 27796859

Werneck AO, Silva DR, Malta DC, Lima MG, Souza-

Júnior PRB, Azevedo LO, Barros MBA, Szwarcwald

CL. The mediation role of sleep quality in the

association between the incidence of unhealthy

movement behaviors during the COVID-19 quarantine

and mental health. Sleep Med. 2020 Dec; 76:10-15. doi:

10.1016/j.sleep.2020.09.021. Epub 2020 Sep 25.

PMID: 33049547; PMCID: PMC7518797.

Werneck AO, Hoare E, Silva DR. Do TV viewing and

frequency of ultra-processed food consumption share

mediators in relation to adolescent anxiety-induced

sleep disturbance? Public Health Nutr. 2021 Jan 27:1-

7. doi: 10.1017/S1368980021000379. Epub ahead of

print. PMID: 33500011.

Yu B, Gu Y, Bao X, Meng G, Wu H, Zhang Q, Liu L, Sun

S, Wang X, Zhou M, Jia Q, Song K, Niu K. Distinct

associations of computer/mobile devices use and TV

watching with depressive symptoms in adults: A large

population study in China. Depress Anxiety. 2019

Sep;36(9):879-886. doi: 10.1002/da.22932. Epub 2019

Jul 3. PMID: 31268210.

ICHIH 2022 - International Conference on Health Big Data and Intelligent Healthcare

416