Abstract
Importance Digital media takes an increasingly large part of children’s time, but the effect on brain development is unclear.
Objective To investigate the effects of digital media (watching television and videos, using social media, or playing video games respectively) on the development of the cortex, striatum, and cerebellum over two years.
Design, setting, and participants A prospective, multicenter, longitudinal study of children from the Adolescent Brain and Cognitive Development Study, recruited between 2016-2018. Children underwent magnetic resonance imaging scan at two different time points and completed the Youth Screen Time Survey at first timepoint, answering questions about digital media use. The analysis controlled for differences in socioeconomic status (SES) and polygenic scores for educational attainment.
Exposure Digital media use.
Main outcome measures The primary outcome measure was the changes in the global cortical surface area.
Results 6492 children (age in months, mean [SD] = 118.6 [7.2], i.e., 9.9 years were included at the baseline. Of these, 4502 children (age in months = 142.6 [7.6], i.e., 11.9 years were included at the two years follow-up. The average time spent by children on screen time was 2.2h/day for watching television and videos, 0.4h/day for using social media, and 0.9h/day for playing video games.
Over the two-year observation period, the average cortical surface area increased by approximately 2%, reflecting normal cortical development. The amount of time spent playing video games was weakly but positively correlated to change in global cortical surface area (standardised beta, β = 0.03; 95% CI [0.001 – 0.06]; P=.06). No global or regional effect on brain development was observed for a time watching television and videos or using social media. However, the regional analysis showed that playing video games was associated with a larger increase in the volume of the cerebellum (β = 0.01 [0.001 – 0.02]; P=.02).
Conclusions and Relevance This study does not suggest that digital media use in children harms brain development in mid-childhood and within a window of two years, but a longer follow-up is necessary.
Question Does the use of digital media affect brain development in children aged 9-10 years of age?
Findings In this two-year longitudinal study of 4502 children, we found no effect of playing video games, watching television, or using social media on the development of cortical surface area. playing video games was associated with a larger increase in the volume of cerebellum.
Meaning This study does not indicate that the use of digital media harms brain development.
Introduction
Children are spending more time with digital media than ever before, but the effects of this on psychological health, cognition, and brain maturation are still unclear. There are growing concerns among parents, caregivers, and policymakers about the negative impact of digital media use on the developing brain of children. In 2016, the American Academy of Paediatrics recommended to limit media use to less than 2h/day for children aged 6-10 years. However, children of this group worldwide, do not meet this recommendation. For example, in the USA, children aged 8-12 years on average spent 4h and 44mins per day on digital media for entertainment purposes,1 aside from use during school and homework. Therefore, studies focusing on the long-term effect of digital media use on children are warranted.
Behavioural studies have found benefits as well as adverse consequences associated with the amount of time spent on digital media activities amongst school-aged children.2-5 For example, longer exposure to digital media use was associated with poor mental health, lower academic performance, and worse behavioural problems in children aged 9-10 years.5
Cross-sectional neuroimaging studies have associated digital media use mainly with smaller grey matter volumes in various cortical regions.6-9 However, these studies fail to account for self-selection. A prior study from our group showed that there are associations between digital media use to both intelligence, SES and genetic predisposition at baseline, but these effects disappear or are sometimes reversed when looking longitudinally, suggesting self-selection bias, highlighting the importance of longitudinal studies. Childhood brain development is affected by environmental characteristics such as upbringing, parental education, parental income, and genetics.10 and these need to be taken into account.
To our knowledge, only one longitudinal study investigated the associations between weekly videogame use and brain changes one-year later in children aged 7-11 years, and found significant changes in functional connectivity between the dorsal putamen and motor cortex,11 but did not control for SES or genetic effects.
To overcome the above limitations, we investigated the effect of digital media use on brain development over two years in children aged 9-10 years at baseline, accounting for age, sex, handedness (to avoid lateralization of the brain structures associated with handedness), SES, scanner sites, and genetics. We separately analyzed different categories of digital media use: watching television and videos, using social media, or playing video games respectively Our main outcome measure was global cortical surface area, which is known to be the measure most highly correlated to intelligence in both children and young adults.12,13 It has also been shown to be relevant for other environmental variables such as SES.14 We also investigated the individual cortical regions and the volumes of striatum and cerebellum which have been implicated in cross-sectional studies of digital media.11,15
Methods
Participants
The neuroimaging and behavioural data used in this study were obtained from the ABCD consortium (data release 4.0 https://abcdstudy.org/; http://doi.org/10.15154/1523041). This longitudinal cohort of 11,875 children was enrolled at baseline (9-10 years) and followed-up two years later, at 21 research sites across the United States.16 Children were excluded if they were born extremely preterm (<28 weeks) or with birth weight (<1200g); were not proficient in English; had any neurological problems; had a history of seizures; or had a contraindication to undergo brain magnetic resonance imaging (MRI) scans. All children and their parents/guardians provided informed written consent/assent for participation, and the central IRB at the UCSD approved the study protocols.
Neuroimaging
Children underwent a 3Tesla brain MRI scan at two time-points using a T1-weighted MP-RAGE sequence (1mm isotropic) which was processed using FreeSurfer (version 5.3.0)–described in detail elsewhere.17 All neuroimaging data were quality controlled by the ABCD consortium. In this study, we used the Destrieux atlas (i.e., parcellates each hemisphere into 74 regions of interest).
Children who did not have relevant data on either SES, genetics, or digital media use were excluded. Additionally, the ABCD cohort included twins and siblings, therefore we randomly selected one child per family to eliminate this source of bias.
Socioeconomic status
SES was defined as the first principal component from a probabilistic principal component analysis of total household income, parental education, and neighbourhood quality. Children missing more than one of these SES measures were excluded. Parental education was categorised into middle school or less (1), some high school (2), high school graduate (3), some college/associates degree (4), bachelor’s degree (5), a master’s degree (6), or professional degree (7). The neighbourhood quality measure was the area deprivation index calculated from the American Community Survey using the address of the primary residency.18 The SES composite and each subcomponent were normalised (mean=0, sd=1).
Polygenic score for cognition
We used a multitrait polygenic score for cognitive performance derived from 1.1 million individuals using their educational attainment, mathematical ability, and general cognitive ability. For more details see eMethods in the Supplement.
Exposures
Digital media use was measured using the Youth Screen Time Survey at baseline visit (i.e., 9-10 years). Children were asked to report the number of hours spent on a typical weekday and weekend days for the following: (1) watching television or movies, (2) watching videos (e.g., YouTube), (3) playing video games on a computer, console, phone, or another device (e.g., Xbox, PlayStation, iPad), (4) Texting on a cell phone, tablet, or computer (e.g., Google Chat, WhatsApp), (5) Visiting social networking sites (e.g., Facebook, Twitter, Instagram), (6) Using video chat (e.g., Skype, FaceTime). Digital media activities were categorised as (a) watching television and videos (1+2), (b) using social media (4+5+6), and (c) playing video games (3). Children responded to each question using a seven-point scale: none, <30min, 30min, 1h, 2h, 3h, to >4h.
Outcomes
We prespecified our primary outcome as a global cortical surface area; secondary outcomes as volumes of striatum (combined caudate nucleus, putamen, and accumbens) and cerebellum (both grey and white matter structures), and individual cortical regions as exploratory outcomes.
Data Analysis
A multiple regression model was used to investigate the association between the individual digital media use and brain regions, both cross-sectionally and longitudinally.
Age at baseline MRI scan visit, sex, 20 principal components from the genetic data, a polygenic score for cognition, SES, handedness, and MRI scanner sites were used as covariates for the cross-sectional analyses (Model 1). For the longitudinal analyses, we performed two models. Model 2, where we investigated the longitudinal effect for each type of digital media use separately (i.e., one statistical model for each type of digital media use), adjusting for differences in age of the child (i.e., age at two years of follow-up MRI scan visit – the age at baseline MRI scan visit) as an additional covariate, in addition to all the previously mentioned covariates in the Model 1. In Model 3, we included all three measures of digital media use in the same model, adjusting for all the covariates as included in Model 2. This model thus corrects for the total amount of digital media used and informs about the specific effect of one activity. We report both standardized beta estimates (mean=0, sd=1) and absolute values for surface area (mm2) and subcortical volumes (mm3). To limit the number of statistical tests, we computed bilateral estimates by combining both the left and right sides of the regions. For the longitudinal analyses, we used the differences in absolute values of brain measures (i.e., two years follow-up –baseline visit).
Analysis of the primary and secondary outcomes was unadjusted for multiple comparisons, as they were a priori. The exploratory outcomes were adjusted for multiple comparisons using the false discovery rate correction. All analyses were carried out using R (4.2.0) with an alpha level of <.05.
Results
Of the 11,875 children recruited into the ABCD study cohort, 6492 children (age in months, mean [SD] = 118.6 [7.2], i.e., 9.9 years; boys, n (%) = 3382 (52%); right-handed, n (%) = 5176 (80%)) fulfilled our inclusion criteria were included at the baseline (i.e., 9-10 years of age). Of these, 4502 children (age in months = 142.6 [7.6], i.e., 11.9 years; boys = 2427 (54%)) were included at the two years of follow-up. The number of children followed up at two years of follow-up was slightly lower compared to the baseline because the COVID-19 pandemic had significantly affected the data collection.
The average time spent by children on digital media use was 2.2h/day for watching television and videos, 0.4h/day for using social media, and 0.9h/day for playing video games.
Primary outcome
At baseline, children who spent more time watching television and videos had a smaller global cortical surface area (Table 1) (standardised beta, β = -0.03; 95% standardised CI, [-0.06 – -0.01]; P=.001). This association was also found for playing video games (β = -0.02 [-0.05 – -0.01]; P=.007), but not using social media (β = -0.002[-0.02 – 0.02]; P=.84).
Association between digital media use and brain regions in children at both baseline and two-years later
At two years of follow-up, the average cortical surface area increased by approximately 2%, reflecting normal cortical development. Children who spent more time playing video games (β=0.03 [0.001 – 0.06]; P=.04) had a larger change in the global cortical surface area. Whereas, children who spent more time watching television and videos (β = -0.01 [-0.03 – 0.02]; P=.59) or using social media (β = -0.002 [-0.02 – 0.03]; P=.92) had no association with the change in global cortical surface area.
Although a small effect was found between playing video games and change in global cortical surface two years later, we were sceptical, as there was a negative association between playing video games and global cortical surface area at baseline, indicating that those who played more had a smaller brain at baseline. This raised two possibilities, whether a significant change in the global cortical surface area two years later is because of normal brain development during this window period, or because of the benefit of playing video games. To check for an unbiased result in our sample (i.e., no association between playing video games and global surface area at baseline), we did a separate subgroup analysis of boys and girls; and children born preterm and at term (Table 2; Figure 1). Previous studies have shown that boys and children born preterm tend to spent more time on digital media.4,19 Consistent with the previous findings, we found that boys spent more time playing video games compared to girls (boys = 1.25 [1.15]; girls = 0.63 [0.84]; Punadj<.001), this was also the case for children born preterm (preterm = 1.02 [1.09]; full-term = 0.94 [1.05]; Punadj=.02). We also did a separate regression analysis (Model 1 and 2) for subgroups and found a negative association between playing video games and global cortical surface area in boys (β = -0.04 [-0.07 – 0.005]; P=.02) and in children born at term (β = -0.03 [-0.05 – 0.007]; Punadj=.01) but not girls (β = -0.02 [-0.06 – 0.007]; Punadj=.13) or children born preterm (β = -0.02 [-0.08 – 0.07]; Punadj=.41) at baseline (eTable 2-5 in the Supplement). During the follow-up, playing video games was positively associated with change in global cortical surface area in boys (β = 0.04 [0.002 – 0.07]; Punadj=.03) and in children born at term (β = 0.03 [0.005 – 0.06]; Punadj=.02) but not girls (β = 0.01 [-0.02 – 0.05]; Punadj=.50) or children born preterm (β = -0.007 [-0.08 – 0.07]; Punadj=.85).
Digital media use and brain measures separated according to sex and gestational length
Box and whisker plots representing the distributions of global cortical surface area (mm2) and cerebellum volume (mm3) across the groups (All, overall cohort; G, girls; B, boys; Term, children born > 37weeks’ of gestation; Preterm, children born < 37weeks’ of gestation). The length of the box represents the interquartile range (IQR), the horizontal line in the box interior represents the median, and the whiskers represent the 1.5 IQR of the 25th quartile or 1.5 IQR of the 75th quartile, and the dots represent outliers.
In an additional attempt to correct for the smaller brain at baseline, we built a model (Models 4 and 5), accounting for the baseline brain measure as an additional covariate to the pre-existing covariates for the longitudinal analyses (as described in the methods). Absolute brain measure at two years of follow-up was used as the dependent variable. Here we found that children who spent more time playing video games at 9-10 years had a trend toward larger global cortical surface area two years later, but did not reach the level of statistical significance (β = 0.006 [-0.0001 – 0.01]; P=.06) (Model 4; Table 1).
Secondary outcomes
At baseline, children who spent more time watching television and videos (β = -0.03 [-0.06 – -0.01]; P=.002), or playing video games (β = -0.03 [-0.05 – -0.01]; P=.01) had a smaller volume of the striatum (Table 1). However, there was no association between watching television and videos, or playing video games with the volume of cerebellum.
For the two years of follow-up, Model 4 and Model 5 were used for analysing the secondary outcome regions for longitudinal analyses (as discussed above).
At two years of follow-up, children who spent more time playing video games had a larger change in the volume of cerebellum (β = 0.01 [0.001 – 0.02]; P=.02) (Model 4; Table 1). However, there was no association between playing video games and change in the volume of striatum. And there was also no association between watching television and videos or using social media and change in the volumes of striatum and cerebellum. The results from the predefined: Model 2 and Model 3 are presented as supplements (eTable 1 in the Supplement).
When all the digital media use (i.e., watching television and videos, using social media, and playing video games) were included in the same model (Model 5), children who spent more time playing video games at 9-10 years of age were independently associated with a larger change in the global cortical surface area (β = 0.008 [0.001 – 0.01]; P=.01) and with a larger change in the volume of cerebellum (β = 0.01 [0.003 – 0.02]; P=.01) at two years later (eTable 1 in the Supplement). However, there was no association between any of the digital media use and the volume of striatum.
Exploratory outcomes
We investigated all three models for effect in each of the individual regions (collapsed across hemispheres). In the regional analyses, there was no association between digital media use and brain regions, after correcting for multiple comparisons (eTable 6-8 in the Supplement).
Discussion
In this large prospective cohort study, we investigated the effects of digital media use on development of cortex, striatum, and cerebellum across two different time points during mid-childhood. Contrary to our hypothesis, the effects of playing video games did not alter the development globally or of any individual cortical region. However, children who spent more time playing video games at 9-10 years had a larger increase in the volume of cerebellum two years later. This association remained significant and stronger when total time spent on digital media use was taken into consideration, suggesting independent effects of playing video games over watching television and videos or using social media in children. Interestingly, we did not find any effect of watching television and videos or using social media on brain development.
In the cross-sectional analysis (Model 1; Table 1), there was a negative association between watching television and videos or playing video games but not using social media with global cortical surface area and striatum volume at 9-10 years of age. This might be evidence of self-selection as reported in our earlier analysis of the same ABCD dataset. There we found a negative association between SES, genetics to the usage of digital media.4
A hotly discussed question is whether exposure to digital media among children benefits or harms the brain.15,20 We did not find any evidence of harmful effects of digital media use. There was a trend for a larger increase in global cortical surface area among heavy gamers. But this could be due to self-selection of playing video games among individuals with smaller, and possibly less mature cortices in combination with prior observations that global cortical surface area during this period of time on average should increase with a peak at approximately at 11 years of age.21 Further, in our post-hoc analysis, we found no association between playing video games and changes in global cortical surface area in either girls or preterm birth, two sub-cohorts where playing video game was not associated with global cortical surface area at baseline. Moreover, the effect of playing video games on global cortical surface area disappeared (Table 1) when we adjusted for the baseline measure (i.e., Model 4) in the overall cohort. These analyses all suggest that the trend observed in the global cortical surface area likely reflects accelerated brain development in children with smaller brains, rather than the positive benefit of playing video games.
We found an effect of playing video games on the volume of the cerebellum, a finding which also held up in subsequent analyses (Model 4). Previous cross-sectional and longitudinal studies have also reported associations between gaming and the volume of the cerebellum 11,15,22 The cerebellum is extensively connected to many regions of the cortex and is suggested to play an important role in planning, inhibition, flexibility, working memory, cognition, and hand-eye coordination.23 Moreover, it is generally involved in cognition, including sequence learning, timing, prediction, and error detection.24 Playing video games requires extraordinary speed, a high degree of perception, a high level of attention, planning, and hand-eye coordination.
The strengths of using the ABCD cohort are a large number of subjects, harmonisation of acquisition protocol over the different sites, a longitudinal design, and including both SES and genetic factors. Our analysis also highlights the importance of correcting for baseline differences which could arise from self-selection. A limitation is that the ABCD study consortium collected the daily digital media use from children using the Youth Screen Time Survey, which is not a quantitative measurement of screen time. It also fails to capture any information about the genre of video games. It is possible that different types of gaming content can differently alter children’s brain development.
Gaming was positively associated with changes in the volume of the cerebellum. No effects were seen for other types of digital media use. These results suggest that digital media use does not affect brain development at least in mid-childhood and in the time frame of two years. Longer follow-ups and other stages of development are necessary, yet our results are promising– digital media use seems not to alter the cortex to any meaningful extent.
Data Availability
All data produced in the present study are available upon reasonable request to the authors
Funding/Support
This study was supported by the Swedish Research Council to Torkel Klingberg
Acknowledgements
Data used in the preparation of this article were obtained from the Adolescent Brain Cognitive Development (ABCD) Study (https://abcdstudy.org), held in the NIMH Data Archive (NDA).
The ABCD study was supported by the National Institutes of Health and additional federal partners under award numbers U01DA041022, U01DA041025, U01DA041028, U01DA041048, U01DA041089, U01DA041093, U01DA041106, U01DA041117, U01DA041120, U01DA041134, U01DA041148, U01DA041156, U01DA041174, U24DA041123, and U24DA041147. A full list of supporters is available at https://abcdstudy.org/?s=nIH+collaborators. A listing of participating sites and a complete listing of the study investigators can be found at https://abcdstudy.org/principal-investigators.html. The ABCD consortium investigators designed and implemented the study and/or provided data but did not necessarily participate in the analysis or writing of this report. This manuscript reflects the views of the authors and may not reflect the opinions or views of the NIH or ABCD consortium investigators.
