Professional Documents
Culture Documents
Acute Effects of Video-Game Playing Versus Television Viewing On Stress Markers and Food Intake in Overweight and Obese Young Men Mario Siervo 2018
Acute Effects of Video-Game Playing Versus Television Viewing On Stress Markers and Food Intake in Overweight and Obese Young Men Mario Siervo 2018
Appetite
journal homepage: www.elsevier.com/locate/appet
a r t i c l e i n f o a b s t r a c t
Article history: Background: Sedentary or near-sedentary activities are associated with overweight/obesity in epide-
Received 12 January 2017 miological studies. This has traditionally been attributed to physical activity displacement. A little-
Received in revised form explored area is whether behavioural stresses alter sensations of appetite and eating behaviour. We
17 August 2017
examined whether behaviours conducted seated (television viewing, video gaming) induce different
Accepted 18 August 2017
Available online 24 August 2017
eating patterns, associated with differential levels of stress response.
Methods and findings: We conducted a randomized controlled trial in 72 overweight/obese adult males,
assigned to three groups (24 per group): (i) non-violent television (control group); (ii) non-violent game
Keywords:
Video game playing
(FIFA); (iii) violent game (Call of Duty). Following a standardized breakfast, the 1-h intervention was
Stress followed by 25-min rest, with sweet and savoury snacks and drinks available ad libitum. Stress markers
Appetite (heart rate, blood pressure, visual analogue scale (VAS)) were measured throughout. Heart rate, systolic
Obesity blood pressure, and stress by VAS were significantly higher (p < 0.05) playing video games than watching
non-violent television, though the two game groups did not differ. Considered separately, only the vi-
olent video game group consumed more energy (D ¼ 208.3 kcal, 95%CI 16, 400), sweet foods (D ¼ 25.9 g,
95%CI 9.9, 41.9) and saturated fat (D ¼ 4.36 g, 95%CI 0.76, 7.96) than controls.
Conclusion: Playing video games in overweight/obese adult males is associated with an acute stress
response relative to watching non-violent television, associated with greater subsequent food intake.
These findings highlight the need to focus on the metabolic effects, as well as the energy costs, of ac-
tivities involving sitting in relation to obesity risk.
© 2017 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.appet.2017.08.018
0195-6663/© 2017 Elsevier Ltd. All rights reserved.
M. Siervo et al. / Appetite 120 (2018) 100e108 101
approach prevents understanding of the metabolic consequences Nieuwenhuizen, Lemmens, Born, & Westerterp-Plantenga, 2009).
of behaviour among adolescents and young adults (eg 18e30 Both of these two studies indicate a tendency to consume foods rich
years), who should be a primary target of obesity prevention in glucose in response to stress (Hitze et al., 2010; Rutters et al.,
campaigns. 2009), as also suggested in a trial of violent video games (Siervo
Sedentary behaviour has been consistently associated with et al., 2013). Finally, a study of brain metabolism demonstrated a
obesity. For example, television viewing is a well-established risk 12% increase in global cerebral metabolic rate for glucose in
factor (Robinson, 2001; Viner & Cole, 2005), and interventions to response to a behavioural stressor (Lund Madsen et al., 1995). This
reduce TV viewing have reduced body mass index (BMI) (Otten, increased glucose utilization persisted for 40min after termina-
Jones, Littenberg, & Harvey-Berino, 2009). Sitting has also been tion of the stressor, despite all physiological stressors having
associated with obesity risk (Hu, Li, Colditz, Willett, & Manson, returned to baseline levels (Lund Madsen et al., 1995).
2003; Kong, Lee, Kim, Sim, & Choi, 2015). Likewise, growth of Collectively, these studies suggest that increased energy intake,
the video games industry has closely paralleled the obesity potentially targeting certain food types, may be an adaptive
epidemic (Wack & Tantleff-Dunn, 2009), and a number of studies response to satisfy the brain's demand for extra cerebral energy
have reported associations between time spent playing video during periods of stress. However, few studies (Chaput et al., 2011;
games and BMI in children (Mota, Ribeiro, Santos, & Gomes, 2006; Hitze et al., 2010; Rutters et al., 2009) have directly measured food
Stettler, Signer, & Suter, 2004), adolescents (Schneider, Dunton, & intake, hence there is little understanding as yet of how video
Cooper, 2007) and adults (Weaver et al., 2009). However, playing games might impact eating behaviour through the stress response.
video games using a hand-held joystick while seated has been This issue can be addressed by probing in greater detail the effect of
associated with greater energy expenditure, similar to mild- different kinds of video games on sensations of appetite and eating
intensity exercise, compared to simply sitting or standing still behaviour.
(Segal & Dietz, 1991). Other studies have reported a modest in- We aimed to examine whether behaviours conducted while
crease in energy expenditure when playing video games compared sitting (television viewing and playing different Playstation video
to resting (Chaput et al., 2011; Mansoubi et al., 2015; Mellecker & games) induce different patterns of eating, associated with differ-
McManus, 2008; Wang & Perry, 2006). On this basis, television ential levels of stress response in young overweight men. Our pri-
viewing would be predicted to be a stronger risk factor for obesity mary hypothesis, specified prior to the trial, was that playing
than playing video games. Recently, a formal definition of seden- violent video games would induce a stress response, an elevation of
tary activity has been proposed of 1.5 METS (Metabolic Equiva- appetite, and a heightened consumption for sweet foods or drinks,
lent Task, defined as 3.5 ml O2$kg1 min1) (Sedentary relative to watching non-violent television or playing a non-violent
Behaviour Research Network, 2012). Studies show that adults video game. However, when analysing the data we also tested two
playing seated Playstation-type video games typically achieve secondary hypotheses. First, we tested whether either of the video
around 1.4 METs, though there is variability between individuals game groups differed from the non-violent television control group
some of which reflects nutritional status assessed by BMI in stress responses, appetite and food consumption. This differs
(Mansoubi et al., 2015). from the primary hypothesis, which not only assumed that the
A less intuitive possibility is that activities while sedentary or violent game would differ from the non-violent television group,
sitting affect appetite, and potentially in variable ways. A ran- but also assumed that the non-violent game would not show such a
domized trial reported higher subsequent energy intake in video- difference from the television group. Second, we also tested the
game players relative to controls, translating into an energy sur- hypothesis that all videogame players combined into a single group
plus of 104.2 kcal over the entire day, despite no reported increases had different stress responses, appetite and food consumption
in hunger or appetite rating (Chaput et al., 2011). Another trial compared to those watching non-violent television. These hy-
found that those playing violent video games on a playstation potheses are illustrated in Supplementary online Fig. S1. We tested
subsequently felt less full and expressed greater desire to eat these hypotheses experimentally using a randomized controlled
sweetened foods, compared to those playing non-violent games or trial, which we undertook in overweight and obese men aged
watching non-violent television (Siervo, Sabatini, Fewtrell, & Wells, 18e30 years, as many regular video gamers are overweight (Wack
2013). More generally, activities involving mental effort (Chaput & & Tantleff-Dunn, 2009).
Tremblay, 2007; Chaput, Drapeau, Poirier, Teasdale, & Tremblay,
2008; Tremblay & Therrien, 2006) and watching television 2. Methods
(Bellisle, Dalix, & Slama, 2004; Temple, Giacomelli, Kent,
Roemmich, & Epstein, 2007) have been reported to increase food A three-arm, parallel, randomized control trial was carried out
intake without elevating appetite. at the Childhood Nutrition Research Centre, University College
It has been postulated that appetite increases following video London (Clinical-Trials.gov NCT01809470). Participants were
gaming may be due to stress-induced rewards or impairment in blinded to the main aim of the study, in order to minimise any
satiety signalling (Chaput et al., 2011). A number of studies have influence of individual expectations. After the study the partici-
documented a stress response after playing video games, shown by pants were fully informed about the outcomes that were
increased heart rate and blood pressure (Barlett & Rodeheffer, measured and monitored. The study was approved by the ‘UCL
2009; Chaput et al., 2011; Segal & Dietz, 1991), while experi- Research Ethics Committee’ (0326/011), and all participants pro-
mental studies have also linked exposure to stress with altered vided written informed consent. All data was anonymized and
patterns of food consumption. A small study (n ¼ 10) found that stored in accordance with the Data Protection Act 1998. The study
exposure to stress significantly increased consumption of carbo- was run from 18th February to 24th May 2013, and finished due to
hydrate, and led to non-significant increases in the consumption of completion. The trial registration was completed 8th March 2013,
sweet foods, fatty foods and total energy (Hitze et al., 2010). due to delays in providing supporting information, but no changes
Another study demonstrated greater total energy intake after to the protocol were made. Funding was provided by University
completing difficult mental arithmetic questions compared to the College London in support of JG. The trial protocol can be accessed
control state (230.6 vs 189.6 kcal, p < 0.01), despite no difference in as supplementary online material (S1 File). The authors confirm
hunger; a statistically significant greater consumption of sweet that all ongoing and related trials for this intervention are
foods was also observed 169.2 vs 143.2 kcal, p < 0.03) (Rutters, registered.
102 M. Siervo et al. / Appetite 120 (2018) 100e108
the R language and environment for statistical computing and individuals, and fixed effects on Time, Group and their interaction
graphics, version 3.0.2 (R: A Language and Environment for showed no differences in SBP, DBP, heart rate or VAS between the
Statistical Computing, 2013). three groups any time point (p > 0.07), hence the violent video
We formally tested normality using the Shapiro-Wilk test. The game group did not show elevated stress relative to the other two
‘food intake’ outcomes did not satisfy this assumption and were groups, our primary hypothesis (Table 1). Regarding secondary
transformed using the Box-Cox family (Box & Cox, 1964) as hypothesis A, also we did not find that either the violent video
implemented in the R library MASS (Venables & Ripley, 2002). A game group or the non-violent video game group differed inde-
square root transformation was judged appropriate based on partial pendently from the non-violent television group by independent
log-likelihood plots; thus we fitted linear regression models samples t-test. However, Table 1 indicated differences between the
including a treatment contrastp with
ffiffiffiffi TV as the reference category to TV and combined Video Game groups in stress markers, hence we
the transformed outcomes ð Y 1Þ=2 , where Y denotes the ran additional mixed effects models to test Secondary hypothesis B.
original food intake variable. For SBP, there were no significant difference in slopes
To compare the three groups individually, baseline characteris- (p ¼ 0.966) nor intercepts (p ¼ 0.068) between TV and combined
tics were compared by one-way ANOVA. We fitted linear mixed- Video Game groups. However, these groups did differ significantly
effects models (Pinheiro & Bates, 2000) with a random effect at the beginning of the resting phase, as the combined Video Game
term in their intercepts to account for the within-participant group was now on average 3.96 mm/Hg higher than the TV Group
repeated measurements. We used maximum likelihood estima- (95% CI 0.50, 7.42, p ¼ 0.025) and both increased significantly at the
tion procedures in order to satisfy the assumptions required to same rate (0.071 (mm/Hg)/min, p ¼ 0.019; LRT for interaction
perform likelihood ratio tests (LRT) in the comparisons of nested p ¼ 0.947) through the rest period, indicating that this difference
models with and without an interaction term between time and was maintained after the intervention (Fig. 1). These SBP data
group identity. therefore support hypothesis 2B. However, no such differences
For the secondary hypotheses, we analysed differences between were found for DBP.
two groups. For secondary hypothesis A, we compared each video For heart rate the TV and combined Video Game groups had no
game group separately against the other group (FIFA vs CoD), and significant trend through the intervention (p ¼ 0.379) but there was
also against the non-violent television control group (FIFA vs TV; a significant difference in their intercepts (p ¼ 0.023) with the
CoD vs TV). For secondary hypothesis B, we combined Groups 2 and combined Video Games groups being 2.71 (95% CI 0.41, 5.01) bpm
3 into a single Video Game group and compared that against the higher on average than the TV group at the beginning of the first
non-violent television control group. As the three hypotheses are phase (Fig. 2). The interaction term between time elapsed and
tested on the same data for stress markers and appetite, we cor- group during the resting phase was not significant (LRT p ¼ 0.620);
rected for multiple comparisons with a Bonferroni procedure by the common slope in this phase was significantly different from
adjusting the level for statistical significance from a ¼ 0.05 to zero (0.221 bpm/min, p < 0.001) as was the difference in intercepts,
a ¼ 0.0167. with participants assigned to the TV group having a lower heart
Initial analyses of the secondary hypotheses indicated that only rate at the end of the rest period (2.96 bpm, 95% CI 0.02, 5.90,
hypothesis 2B was supported for some stress markers, and also p ¼ 0.047). This therefore supported our secondary hypothesis B,
food intake. For this hypothesis, we additionally fitted three types that the combined video games group developed elevated heart
of mixed-effects models to the stress marker outcomes. Model 1 rate compared to the TV group during the intervention.
was fitted to measurements up to 60 min; it included an intercept For VAS stress, fitting mixed-effects models including a random
for each group and a common slope. Models 2 (no interaction effect on the individuals, and fixed effects on Time, Group and their
time:group) and 3 (including the interaction time:group) were interaction showed differences between each of the VG groups and
fitted to measurements between 60 and 85 min to analyze changes the non-violent TV controls (p ¼ 0.0244 for CoD and p ¼ 0.003 for
between the end of the TV/game intervention and the end of the FIFA), However, these findings did not support our secondary hy-
resting phase. Comparisons between these two groups were con- pothesis A, since the two video game groups did not differ from
ducted using independent t-tests. each other (p > 0.13). Instead, the data again supported secondary
hypothesis B, showing a contrast between the combined video
3. Results games group and the TV group confirmed by independent samples
t-test (Table 1; Fig. 3). The difference in the intercepts was nearly
A total of 477 recruits were assessed for eligibility (Supple- significant (7.20, 95% CI -0.18, 14.58, p ¼ 0.055), indicating that
mentary online Fig. S2). Seventy-two (24 per group) were subse- although similar at baseline, the two groups differed in VAS stress
quently recruited and randomized. Due to small differences during the intervention, though this difference did not steadily
between reported weight and height values, and those measured at increase over the 60-min session (p ¼ 0.183). At the start of the
the time of recruitment, a minority of individuals of each group had resting phase, the intercepts for the TV and combined Video Game
BMI below 25 kg/m2. The median and range were as follows: Group groups still differed significantly, with the combined Video Game
1e25.8, range 24.6e32.9; Group 2e26.4, range 24.5e29.7; Group group having a much higher stress rating (6.75, 95% CI 2.65, 10.85,
3e27.0, range 24.3e33.3 kg/m2. Baseline characteristics did not p ¼ 0.002) and a significant negative slope (0.318 VAS units/min,
statistically differ between groups by one-way ANOVA (Supple- p ¼ 0.012), whilst the slope for the TV group did not differ signif-
mentary online Table S1). icantly from zero (0.011, p ¼ 0.914). A t-test showed that both
Two participants did not strictly adhere to the study protocol. groups ended at similar levels of stress after the resting period
One individual refused to eat the standardized breakfast, and was (p ¼ 0.625), showing that the elevated VAS stress of the video-game
therefore given an alternative meal with similar calorie content. players relative to the television controls rapidly disappeared when
Another participant arrived at the study hung-over, and was the game ended.
probably dehydrated. Analysis was therefore repeated excluding
these data. The significance of the results was not altered by their 3.1. Food intake and desire for specific food types
inclusion/exclusion, and the tables and figures therefore refer to the
full sample. No harm or adverse effect occurred during the study. There were no differences between any group in subjective
Fitting mixed-effects models including a random effect on the hunger rating, or subjective desire to eat any particular food type,
104 M. Siervo et al. / Appetite 120 (2018) 100e108
Table 1
Markers of stress by timepoint for each group.
Group 1 (TV) Group 2 (FIFA) Group 3 (CoD) Group 2 þ 3 Group Group Group
Mean, SD Mean, SD Mean, SD Mean, SD 1 vs 2 1 vs 3 1 vs (2 þ 3)
p-value p-value p-value
SBP (mmHg) Baseline 116.4, 8.6 115.6, 8.6 117.9, 8.1 116.7, 8.3 0.7 0.5 0.8
15 min 114.7, 9.1 115.6, 9.4 118.3, 11.4 116.9, 10.4 0.7 0.2 0.2
30 min 114.9, 10.9 117.1, 10.0 118.6, 12.4 117.8, 11.2 0.4 0.2 0.1
45 min 113.3, 10.1 117.2, 10.4 117.8, 10.5 117.5, 10.3 0.2 0.1 0.032
60 min 113.5, 11.0 117.1, 9.0 118.6, 11.5 117.8, 10.2 0.2 0.1 0.024
85 min 115.4, 9.5 119.8, 9.9 119.3, 9.4 119.7, 9.6 0.1 0.1 0.022
DBP (mmHg) Baseline 65.2, 7.7 64.6, 5.9 67.2, 7.4 65.4, 7.8 0.7 0.3 0.7
15 min 61.4, 7.8 63.3, 6.0 65.8, 9.4 64.1, 8.7 0.3 0.085 0.1
30 min 61.9, 6.7 64.8, 7.0 65.3, 9.2 64.7, 8.7 0.1 0.1 0.1
45 min 62.0, 9.8 63.8, 6.8 65.8, 7.2 64.5, 8.7 0.4 0.1 0.3
60 min 62.5, 6.8 64.5, 6.5 66.2, 6.7 65.2, 7.1 0.3 0.069 0.1
85 min 62.7, 7.8 64.9, 6.2 66.1, 8.8 65.0, 8.3 0.3 0.17 0.2
HR (bpm) Baseline 75.2, 10.6 74.6, 9.8 72.9, 9.7 73.7, 9.7 0.8 0.4 0.5
15 min 69.0, 11.3 71.7, 9.2 69.7, 8.7 70.7, 8.9 0.3 0.8 0.002
30 min 71.3, 11.6 73.2, 9.5 70.2, 7.8 71.7, 8.8 0.5 0.6 0.1
45 min 70.9, 10.9 74.2, 9.1 70.2, 7.4 72.2, 8.4 0.2 0.8 0.052
60 min 69.7, 10.8 72.8, 9.3 69.3, 7.5 71.1, 8.6 0.3 0.8 0.025
85 min 75.7, 9.7 77.7, 11.2 75.1, 7.8 76.5, 9.7 0.5 0.8 0.086
VAS (stress) Baseline 13.6, 13.7 11.8, 10.5 16.4, 18.6 14.2, 15.2 0.6 0.5 0.8
15 min 9.6, 7.6 17.0, 16.5 21.0, 14.4 19.5, 15.4 0.030 0.001 0.009
30 min 7.4, 6.0 18.2, 25.3 25.4, 17.7 21.8, 16.5 0.002 0.001 0.001
45 min 9.4, 8.4 16.4, 14.5 22.4, 19.1 19.4, 17.0 0.045 0.004 0.031
60 min 7.8, 6.8 16.1, 13.2 21.3, 20.2 18.7, 17.1 0.009 0.004 0.018
85 min 7.5, 6.3 9.3, 7.4 11.2, 9.93 10.3, 8.7 0.3 0.1 0.5
SBP e systolic blood pressure; DBP e diastolic blood pressure; HR e heart rate; bpm e beats per minute; VAS e Visual Analogue Scale; d e change from baseline; SD e standard
deviation; SE e standard error. Comparisons made by a t-test for independent samples.
Fig. 1. Change in systolic blood pressure (SBP) from baseline for the Television and
Fig. 2. Change in heart rate from baseline in Television and combined Video Games
combined Video Games groups (Hypothesis 2B). Data are mean and standard error.
groups (Secondary hypothesis B). Data are mean and standard error. The groups
The groups achieved a difference during the intervention (D ¼ 3.96 mm/Hg, p ¼ 0.025)
achieved a difference during the intervention (D ¼ 2.71 bpm, p ¼ 0.023) that persisted
that was maintained during the rest period.
through the rest period.
4. Discussion
compared to non-violent video games in normal weight adult men intakes, only the violent game group was significantly different to
(Siervo et al., 2013). Second, it is possible that in response to violent the TV group for these outcomes, consuming significantly more
video gaming, a different and larger cerebral stress response is energy (37.4%), sugar (70.2%), added sugar (79.6%) and saturated fat
taking place, but the physiological markers of stress that we (61.6%) as a percentage of the intake of the TV-watching controls.
measured did not detect this. Third, the level of violence in the Call These outcomes were observed despite a lack of differences in the
of Duty game may have been insufficient (our previous study used desire to eat any specific food type indicated by VAS. Moreover,
‘Grand Theft Auto’ as the violent game (Siervo et al., 2013)), or though not quite significant (p ¼ 0.072), the violent video game
participants did not become sufficiently involved in the game, to group consumed on average 16.0 g (SE 8.7 g) more sugar compared
elicit a greater stress response compared to playing FIFA. to the non-violent game group. Our findings support hypothesis 2A
Whilst our own interest is in the relationship between the stress more strongly than 2B, but the study merits replication with a
response and appetite perturbation, our findings also have clinical larger sample size, to confirm whether violent video games do
significance in terms of the cardiovascular reactivity hypothesis, indeed elevate energy consumption above the level induced by
which assumes that psychological stressors promote the develop- non-violent games.
ment of cardiovascular disease. Repetitive elevations of SBP These findings support the notion that central metabolic path-
through chronic video game playing in the absence of physical ways, elevated by the stress response, may drive greater energy
activity may affect long-term regulation of BP, potentially leading to intake following stress exposure. Previous work showed that an
hypertension (Phillips & Hughes, 2011). increase in cerebral glucose metabolism persisted for 40 min after
cessation of a psychosocial stressor (Lund Madsen et al., 1995). Both
4.2. Food intake fatty and sweet food types can enhance mood and mitigate the
effects of stress through neurotransmission of dopamine and opi-
Playing the violent game for 1 h was accompanied by a higher oids (Davis, Strachan, & Berkson, 2004; Leigh Gibson, 2006). This is
energy intake of 208.3 kcal relative to the non-violent TV group consistent with the observed increase in intake (due to both sweet
(p < 0.03), however no significant difference from the TV group was and savoury foods) shown by the combined video games groups
seen in those playing the non-violent game. Combining the two during the 25 min rest period. The elevated consumption of sweet
groups, energy intake was 172.6 kcal (95%CI 23.3, 321.9, p < 0.04) foods in the violent game group may represent an attempt to rectify
greater than in TV controls, without any apparent increase in higher levels of brain glucose depletion. Chronic exposure to video
feelings of hunger as measured by VAS. Energy expenditure was not gaming might thus lead to overeating of unhealthy foods to alle-
measured directly, however heart rate declined significantly less in viate stress via the reward system, but further work is required to
the Video Games groups than controls. Chaput and co-workers test whether violent and non-violent video games activate the
(Chaput et al., 2011) reported that playing FIFA e one of the ex- stress response differently.
posures used in this study - resulted in a 21.3 kcal/h increase in
energy expenditure during the 1-h intervention. Assuming a 4.3. Strengths, limitations and future studies
similar increase in energy expenditure in our study, the Video
Games group would have demonstrated a net energy gain of The strengths of this study its experimental design, the rela-
151.3 kcal. tively large sample size, and the high level of protocol compliance.
Overall, our findings show consistency with those of Chaput and Several markers of the stress response were obtained, and we
colleagues (Chaput et al., 2011) who demonstrated an increased directly measured actual food intakes, rather than expressions of
caloric intake of 80.1 kcal without any increase in sensations of hunger alone.
hunger in response to 1 h of non-violent video gaming. The overall However, the study also had several limitations. First, a minority
effect of video gaming was greater in our own study (172.6 kcal), of members of each group had BMI <25 kg/m2, and so were not
which might be due to any of several differences between the overweight according to the standard definition. However, the
studies. First, ad libitum intake was tested using spaghetti bolo- differences were relatively small and applied to each of the three
gnese by Chaput et al. (Chaput et al., 2011), whereas we offered a groups, hence this issue is unlikely to have affected our tests of
selection of energy-dense sweet and savoury snacks. Second, the differences between groups. Conversely, the results may not be
average BMI of participants in our study was much greater (27.0 kg/ generalizable to the whole population, as this study was only car-
m2 vs 21.5 kg/m2), which could affect food consumption in the ried out on overweight/obese males. Although the randomized
immediate period after the intervention (though high BMI might design of our study aimed to avoid confounding by potential dif-
also affect energy balance through reducing physical activity level ferences in factors such as habitual/recent energy/macronutrient
later in the day). Third, our controls watched non-violent TV, intake, intake of caffeine or alcohol consumption and physical ac-
whereas those of Chaput et al. were at rest. Fourth, we investigated tivity/sleep patterns that might influence appetite responses to
both non-violent and violent video games, whereas Chaput ana- stress events or activities, further studies could address such factors
lysed only a non-violent video game. Unlike Chaput's study, we did in more detail. We did not strictly control fasting, and accepted the
not find that the FIFA was associated with a significantly greater participants’ word that they had not consumed food since the prior
energy intake than controls, however the study of Chaput et al. had evening. However, the randomized study design should minimise
greater statistical power, due to the within-person cross-over the introduction of artefacts from such variability.
design on 22 participants. Nevertheless, both studies provide evi- Third, we did not take into account differences in the ‘scores’ of
dence in support of the hypothesis that, if not compensated by the video game players, and do not know whether some became
reduced energy intake later in the day, regular appetite perturba- frustrated and stressed because of their own limitations. Fourth,
tions induced by video gaming could play a role in chronic weight our intervention was in the morning and lasted 1 h, whereas many
gain. gamers play for much longer, and often at night-time (Bavelier
The key novel finding in our study was that the presence of et al., 2011; Cummings & Vandewater, 2007). It would be ideal to
violence in the video game appeared to elevate the increase in food measure energy intake and expenditure over a 24-h period to
intake by video game players, relative to watching TV. Although the calculate net daily energy consumption. Fifth, the majority of the
two video game groups did not statistically significantly differ from selections of foods available were unhealthy (though fruit was
each other in relation to energy intake, or specific macronutrient offered), therefore selecting greater quantities of sugar and fat in
M. Siervo et al. / Appetite 120 (2018) 100e108 107
the video game groups did not reflect free choice. Sixth, the dif- Funding/Support
ferences observed in the objective stress markers between control
and video game groups may reflect a relaxing effect in controls, Core budget, Childhood Nutrition Research Centre, UCL Great
rather than increased stress in the video game groups. Lastly, it was Ormond Street Institute of Child Health. Research at the UCL
not possible to conduct this study blind due to logistical reasons; Institute of Child Health and Great Ormond Street Hospital for
therefore the results are potentially subject to bias, whereby sub- Children receives a proportion of the funding from the Department
conscious actions by the investigator could influence the results. To of Health's National Institute for Health Research Biomedical
minimise this possibility, all interactions with participants were Research Centres funding scheme.
standardized in the protocol.
Future studies should include larger sample sizes to increase Clinical trial registration number
statistical power, and should assess central metabolic pathways to
provide stronger evidence of effects on appetite regulation. Clinical-Trials.gov NCT01809470.
Furthermore, it would be valuable to test for training effects of
playing video games, as this would provide better evidence of how Conflict of interest statement
regular game-playing might affect BMI. Finally, such studies should
be conducted on adolescents, who constitute a large part of the The authors have no conflict of interest to declare.
video-game playing population.
Appendix A. Supplementary data
5. Conclusions
Supplementary data related to this article can be found at http://
The conventional approach to the association between seden- dx.doi.org/10.1016/j.appet.2017.08.018.
tary or near-sedentary activities and obesity has been to assume
that they increase risk by reducing energy expenditure, and References
therefore promote positive energy balance. This approach cannot
discriminate specific behaviours that have particular relevance for Barlett, C. P., & Rodeheffer, C. (2009). Effects of realism on extended violent and
nonviolent video game play on aggressive thoughts, feelings, and physiological
obesity risk, with the risk that public health policies based on this arousal. Aggressive Behavior, 35(3), 213e224.
approach may be weak and ineffective. This is particularly relevant Bavelier, D., Green, C. S., Han, D. H., Renshaw, P. F., Merzenich, M. M., & Gentile, D. A.
to adolescents and young adults, who may rapidly develop strong (2011). Brains on video games. Nature Review Neuroscience, 12(12), 763e768.
Bellisle, F., Dalix, A. M., & Slama, G. (2004). Non food-related environmental stimuli
behavioural patterns during this formative period (Falbe et al., induce increased meal intake in healthy women: Comparison of television
2014). Recently, alternative models of obesity causation have viewing versus listening to a recorded story in laboratory settings. Appetite,
been developed, suggesting that specific activities may perturb 43(2), 175e180.
Box, G. E., & Cox, D. R. (1964). An analysis of transformations. Journal of the Royal
cellular metabolic regulation, and hence perturb appetite and Statistical Society, Series B, 26, 211e252.
eating behaviour (Davis et al., 2004; Lustig, 2008; Spiegel, Tasali, Chaput, J.-P., Drapeau, V., Poirier, P., Teasdale, N., & Tremblay, A. (2008). Glycemic
Penev, & Van Cauter, 2004; Wells & Siervo, 2011). Consistent with instability and spontaneous energy intake: Association with knowledge-based
work. Psychosomatic Medicine, 70(7), 797e804.
others (Chaput et al., 2011), our study supports this newer Chaput, J.-P., & Tremblay, A. (2007). Acute effects of knowledge-based work on
approach, showing that compared to watching non-violent televi- feeding behavior and energy intake. Physiological Behaviour, 90, 66e72.
sion, playing video games generates a stress response, and induces Chaput, J.-P., Visby, T., Nyby, S., Klingenberg, L., Gregersen, N. T., Tremblay, A., et al.
(2011). Video game playing increases food intake in adolescents: A randomized
an increased consumption of calories. A novel finding, supporting
crossover study. American Journal of Clinical Nutrition, 93(6), 1196e1203.
our primary hypothesis, was that playing a violent game was Cummings, H. M., & Vandewater, E. A. (2007). Relation of adolescent video game
associated with a significantly greater consumption of sweet foods/ play to time spent in other activities. Archives of Pediatrics and Adolescent
drinks than watching non-violent TV, whereas playing a non- Medicine, 161, 684e689.
Davis, C., Strachan, S., & Berkson, M. (2004). Sensitivity to reward: Implications for
violent game did not generate such a difference. However, unlike overeating and overweight. Appetite, 42, 131e138.
in a previous study (Siervo et al., 2013), we did not confirm our Day, S. J., & Graham, D. F. (1991). Sample size estimation for comparing two or more
primary hypothesis in relation to stress markers, since the violent treatment groups in clinical trials. Statistics in Medicine, 10, 33e43.
De Mey, C., Enterling, D., & Meineke, I. (1993). Cardiovascular effects of eating,
video game group did not show a greater stress response than the atenolol and their interaction: beta1-adrenergic modulation does not play a
other groups. Our study suggests that whilst video games per se predominant role in the genesis of postprandial effects. British Journal of Clinical
may perturb appetite, violent video games may exacerbate this Pharmacology, 36, 427e435.
de Mey, C., Enterling, D., Brendel, E., & Meineke, I. (1987). Postprandial changes in
effect. Our work therefore identifies violent video games as a topic supine and erect heart rate, systemic blood pressure and plasma noradrenaline
meriting further research in relation to obesity prevention and and renin activity in normal subjects. European Journal of Pharmacology, 32,
management. 471e476.
Fagan, T. C., Sawyer, P. R., Gourley, L. A., Lee, J. T., & Gaffney, T. E. (1986). Postprandial
alterations in hemodynamics and blood pressure in normal subjects. American
Journal of Cardiology, 58, 636e641.
Author contributions Falbe, J., Willett, W. C., Rosner, B., Gortmaker, S. L., Sonneville, K. R., & Field, A. E.
(2014). Longitudinal relations of television, electronic games, and digital ver-
Study concept and design: Siervo, Fewtrell, Wells. Acquisition of satile discs with changes in diet in adolescents. American Journal of Clinical
Nutrition, 100(4), 1173e1181.
data: Gan. Analysis and interpretation of data: Siervo, Wells, Gan. Goran, M. I., & Treuth, M. S. (2001). Energy expenditure, physical activity, and
Drafting of the manuscript: Siervo, Gan and Wells. Critical revision obesity in children. Pediatric Clinics of North America, 48(4), 931e953.
of the manuscript for important intellectual content: Siervo, Gan, Hess, G. (1838). The evolution of heat in multiple proportions. Poggendorf Ann
Chemie Physik, 47, 210e212.
Cortina-Borja, Fewtrell, Wells. Statistical analysis: Cortina-Borja. Hitze, B., Hubold, C., van Dyken, R., Schlichting, K., Lehnert, H., Entringer, S., et al.
Obtained funding: Wells. Study supervision: Siervo, Wells. (2010). How the selfish brain organizes its supply and demand. Frontiers of
Neurogenetics, 2, 1e13.
Hu, F. B., Li, T. Y., Colditz, G. A., Willett, W. C., & Manson, J. E. (2003). Television
Financial disclosure watching and other sedentary behaviors in relation to risk of obesity and type 2
diabetes mellitus in women. Journal of American Medical Association, 289(14),
1785e1791.
None reported. Kong, I. G., Lee, H. J., Kim, S. Y., Sim, S., & Choi, H. G. (2015). Physical activity, study
108 M. Siervo et al. / Appetite 120 (2018) 100e108
sitting time, leisure sitting time, and sleep time are differently associated with Segal, K. R., & Dietz, W. H. (1991). Physiologic responses to playing a video game.
obesity in Korean adolescents: A population-based study. Medicine, 94(44), American Journal of Diseases of Children, 145(9), 1034e1036.
e1965. Siervo, M., Sabatini, S., Fewtrell, M. S., & Wells, J. C. K. (2013). Acute effects of violent
Leigh Gibson, E. (2006). Emotional influences on food choice: Sensory, physiological video-game playing on blood pressure and appetite perception in normal-
and psychological pathways. Physiological Behaviour, 89, 53e61. weight young men: A randomized controlled trial. European Journal of Clinical
Lund Madsen, P., Hasselbalch, S. G., Hagemann, L. P., Skovgaard Olsen, K., Bülow, J., Nutrition, 67(12), 1322e1324.
Holm, S., et al. (1995). Persistent resetting of the cerebral oxygen/glucose up- Spiegel, K., Tasali, E., Penev, P., & Van Cauter, E. (2004). Brief communication: Sleep
take ratio by brain Activation: Evidence obtained with the KetyeSchmidt curtailment in healthy young men is associated with decreased leptin levels,
technique. Journal of Cerebral Blood Flow and Metabolism, 15, 485e491. elevated ghrelin levels, and increased hunger and appetite. Annals of Internal
Lustig, R. H. (2008). Which comes first? The obesity or the insulin? The behavior or Medicine, 141(11), 846e850.
the biochemistry? Journal of Pediatrics, 152(5), 601e602. Stettler, N., Signer, T. M., & Suter, P. M. (2004). Electronic games and environmental
Mansoubi, M., Pearson, N., Clemes, S. A., Biddle, S. J., Bodicoat, D. H., Tolfrey, K., et al. factors associated with childhood obesity in Switzerland. Obesity Research,
(2015). Energy expenditure during common sitting and standing tasks: Exam- 12(6), 896e903.
ining the 1.5 MET definition of sedentary behaviour. BMC Public Health, 15, 516. Stubbs, R. J., Hughes, D. A., Johnstone, A. M., Rowley, E., Reid, C., Elia, M., et al.
Mellecker, R. R., & McManus, A. M. (2008). Energy expenditure and cardiovascular (2000). The use of visual analogue scales to assess motivation to eat in human
responses to seated and active gaming in children. Archives of Pediatrics and subjects: A review of their reliability and validity with an evaluation of new
Adolescent Medicine, 162(9), 886e891. hand-held computerized systems for temporal tracking of appetite ratings.
Mota, J., Ribeiro, J. C., Santos, M. P., & Gomes, H. (2006). Obesity, physical activity, British Journal of Nutrition, 84, 405e415.
computer use, and TV viewing in Portuguese adolescents. Pediatric Exercise Temple, J. L., Giacomelli, A. M., Kent, K. M., Roemmich, J. N., & Epstein, L. H. (2007).
Science, 18, 113e121. Television watching increases motivated responding for food and energy intake
Ng, M., Fleming, T., Robinson, M., Thomson, B., Graetz, N., Margono, C., et al. (2014). in children. American Journal of Clinical Nutrition, 85(2), 355e361.
Global, regional, and national prevalence of overweight and obesity in children Tremblay, A., & Therrien, F. (2006). Physical activity and body functionality: Im-
and adults during 1980-2013: A systematic analysis for the global burden of plications for obesity prevention and treatment. Canadian Journal of Physiology
disease study 2013. Lancet. May 28. E-pub ahead of print. and Pharmacology, 84, 149e156.
Otten, J. J., Jones, K. E., Littenberg, B., & Harvey-Berino, J. (2009). Effects of television Venables, W. N., & Ripley, B. D. (2002). Modern applied Statistics with S-PLUS. New
viewing reduction on energy intake and expenditure in overweight and obese York: Springer.
adults: A randomized controlled trial. Archives of Internal Medicine, 169(22), Viner, R. M., & Cole, T. J. (2005). Television viewing in early childhood predicts adult
2109e2115. body mass index. Journal of Pediatrics, 147(4), 429e435.
Phillips, A. C., & Hughes, B. M. (2011). Introductory paper: Cardiovascular reactivity Wack, E., & Tantleff-Dunn, S. (2009). Relationships between electronic game play,
at a crossroads: Where are we now? Biological Psychology, 86, 95e97. obesity, and psychosocial functioning in young men. Cyberpsychology &
Pinheiro, J. C., & Bates, D. M. (2000). Mixed-effects models in S and S-PLUS. New York: Behavior : The Impact of the Internet, Multimedia and Virtual Reality Onbehavior
Springer. and Society, 12(2), 241e244.
R: A Language and Environment for Statistical Computing (2013). Vienna, Austria: R Wang, X., & Perry, A. C. (2006). Metabolic and physiologic responses to video game
Foundation for Statistical Computing. play in 7- to 10-year-old boys. The Archives of Pediatrics & Adolescent Medicine,
Robinson, T. N. (2001). Television viewing and childhood obesity. Pediatric Clinics of 160(4), 411e415.
North America, 48(4), 1017e1025. Weaver, J. B., 3rd, Mays, D., Sargent Weaver, S., Kannenberg, W., Hopkins, G. L.,
Rutters, F., Nieuwenhuizen, A. G., Lemmens, S. G. T., Born, J. M., & Westerterp- Eroglu, D., et al. (2009). Health-risk correlates of video-game playing among
Plantenga, M. S. (2009). Acute stress-related changes in eating in the absence of adults. American Journal of Preventative Medicine, 37(4), 299e305.
hunger. Obesity, 17, 72e77. Weinsier, R. L., Hunter, G. R., Heini, A. F., Goran, M. I., & Sell, S. M. (1998). The eti-
Schneider, M., Dunton, G. F., & Cooper, D. M. (2007). Media use and obesity in ology of obesity: Relative contribution of metabolic factors, diet, and physical
adolescent females. Obesity, 15(9), 2328e2335. activity. American Journal of Medicine, 105(2), 145e150.
Sedentary Behaviour Research Network. (2012). Letter to the Editor: Standardized Wells, J. C. K., & Siervo, M. (2011). Obesity and energy balance: Is the tail wagging
use of the terms “sedentary” and “sedentary behaviours”. Applied Physiology the dog? European Journal of Clinical Nutrition, 65(11), 1173e1189.
Nutrition and Metabolism, 37, 540e542.