Background: The use of digital technology such as mobile phones is ubiquitous in adolescents. However, excessive use may have adverse health effects, possibly partially mediated by disruptions to sleep.
Objective: This study aims to assess the social predictors of digital technology use and their cross-sectional association with BMI z scores and being overweight in a large sample of adolescents.
Methods: We used baseline data from a subset of a large adolescent cohort from 39 schools across Greater London who participated in the Study of Cognition, Adolescents and Mobile Phones (n=1473). Digital technology use included phone calls, internet use on mobile phones, and video gaming on any device. Multilevel regression was used to assess the associations between digital technology use and age-specific and sex-specific BMI z scores and being overweight (including obesity). Measurements were derived from height and weight, obtained by the Tanita BC-418 Body Composition Analyzer. We examined whether these associations were mediated by insufficient sleep.
Results: Generally, participants with lower socioeconomic status reported more use of digital technology. Controlling for socioeconomic status, internet use on mobile phones for more than 3 hours per day was associated with higher BMI z scores (adjusted β=.30, 95% CI 0.11-0.48) and greater odds of being overweight (adjusted odds ratio 1.60, 95% CI 1.09-2.34), compared with low use (≤30 minutes). Similar associations were found between video gaming and BMI z scores and being overweight. The BMI z score was more strongly related to weekday digital technology use (internet use on mobile phones and video gaming) than weekend use. Insufficient sleep partly mediated the associations between digital technology use and BMI z scores (proportion of mediation from 8.6% to 17.8%) by an indirect effect.
Conclusions: We found an association between digital technology use and BMI in adolescents, partly mediated by insufficient sleep, suggesting that the underlying mechanisms may be multifactorial. Further research with longitudinal data is essential to explore the direction of the relationships.
With advances in information and communication technologies, mobile phone use is ubiquitous, especially in young people, with 44% of 8- to 11-year olds and 86% of 12- to 15-year olds in the United Kingdom owning a mobile phone . Adolescents aged 12-15 years also spend an average of 12 hours per week on video gaming [ ]. Digital technology offers a broad range of functions, facilitating access to the internet and social networks, instant and text message exchange, multimedia, and entertainment. Despite unambiguous advantages such as easy access to information and fast communication, overuse of digital technology is prevalent in children and adolescents because of its powerful potential for diversion and escape from daily life [ , ]. This is becoming a major public health concern with adverse effects on physical health, psychological well-being, and academic achievement in adolescents [ - ].
Despite the exponential increase in digital technology use, the digital divide because of socioeconomic disparity is persistent. However, the nature of the divide today is a gap in the type and intensity of digital technology use, rather than a gap in access to digital devices . Specifically, although family income is not associated with mobile phone ownership in adolescents [ ], those with higher family income and parental education are less likely to show problematic mobile phone use or internet addiction as their digital behaviors (eg, time spent on digital devices and digital safety) may be supervised by their parents to a greater extent [ - ]. Adolescents who are marginalized or from disadvantaged groups tend to use digital technology at a higher rate, possibly because they are likely to experience dissatisfaction with their offline lives and web spaces can facilitate the development of relationships with peers with similar life experiences [ , ]. They may also have fewer opportunities to carry out extracurricular activities than their affluent peers and thus spend more time on digital devices [ ]. However, there is a paucity of large-scale studies with participants from diverse socioeconomic backgrounds and nuanced exposure measurements to decipher the digital divide regarding various digital technology use, such as phone calls, internet use, and video gaming.
Previous studies have shown that longer duration of phone calls, internet use on mobile phones, and video gaming are associated with a higher risk of overweight and obesity in adolescents [- ]. However, these studies have small sample sizes [ ], do not specify weekday and weekend use [ , ], and only rely on self-reported height and weight, which is not reliable, especially for adolescents whose growth is rapidly changing [ , , ]. Such limitations preclude a rigorous investigation of the relationship between digital technology use and obesity as well as the underlying mechanism, which is still unclear. The duration of screen-based device use, such as television, mobile phone, and video game consoles, is associated with insufficient sleep duration and increased sleep disturbance in adolescents [ - ]. Possible mechanisms include sleep time displacement by digital technology use and melatonin suppression at bedtime by light exposure emitted from these devices [ ]. Insufficient sleep is a well-documented risk factor for obesity in adolescents through the development of insulin resistance, sedentariness, and unhealthy diet in terms of higher consumption of energy-dense foods and sweetened beverages [ ]. Therefore, exploration of the complex interrelationships between digital technology use, sleep, and obesity in adolescents is crucial for understanding and preventing obesity.
We have recently established the Study of Cognition, Adolescents and Mobile Phones (SCAMP), the largest adolescent cohort study to investigate the use of mobile phones and other wireless devices and its association with cognitive, behavioral, health, and educational outcomes. We previously observed in this cohort that adolescents who used their mobile phones at nighttime, particularly in darkness, had poorer sleep outcomes compared with nonusers .
In this study, we aim to examine the associations between digital technology use (eg, phone calls, internet use on mobile phones, and video gaming) and measured BMI outcomes. In addition, we investigated whether insufficient sleep mediated these associations. We hypothesized that increasing digital technology use is associated with higher BMI z scores and being overweight and that the association between digital technology use and BMI z scores is mediated by insufficient sleep.
Study Design and Participants
SCAMP is a prospective adolescent cohort study, with details reported elsewhere . Baseline data were collected from 6616 year 7 pupils (aged 11-12 years), recruited between 2014 and 2016, from 39 secondary schools (26 state and 13 independent) across Greater London, UK. Of all year 7 pupils (N=7375) at these 39 schools, 111 (1.51%) parents or pupils chose to opt out. The remaining nonparticipation (n=648) was because of absence, nonassents by parents, withdrawals, technical issues, or miscellaneous reasons.
Participants in all SCAMP schools completed a computer-based assessment using Psytools software (Delosis Ltd) in examination mode. The assessment included a questionnaire on their digital technology behaviors (eg, smartphone use, social media engagement, and video gaming); a battery of cognitive tests; and health, well-being, and behavior scales. A subset of SCAMP participants (12 out of 39 schools, n=2270) also participated in SCAMP Bio-Zone to provide noninvasive biological samples (urine and saliva) and anthropometric measurements (height, weight, waist circumference, grip, and pinch strength) as well as perform a lung function test. The SCAMP Bio-Zone participants with measured height and weight (n=1473) were included into the present analysis for assessing the association between digital technology use and BMI. Our sample (n=1473) was powered to detect a small effect (effect size=0.005) for a two-tailed linear regression with 80% power. Our sample size is much larger than the required sample size (n=462) to detect a small mediation effect (eg, β=.14) with 80% power .
Exposure—Digital Technology Use
Categorical responses were provided for questions on digital technology use, for weekdays and weekends separately. Participants who reported using or having used a mobile phone were further asked about the duration of phone calls and internet use (eg, surfing the internet, WhatsApp, Facebook, YouTube, and any other web-based apps) on their mobile phones if their phones were able to connect to the internet. We examined both phone calls and internet use on mobile phones, given the vast majority of mobile phone owners in our study (5119/5490, 93.24%) had a mobile phone that connected to the internet. Internet use on tablets or laptops was not included, as it was not specifically measured. Participants were also asked about the duration of playing video games on any device. We integrated weekday and weekend use to yield average daily use, and categorized internet use on mobile phones and video gaming as 0-30 minutes, 31-59 minutes, 1-2 hours, and 3 hours or more, to allow for a sufficient sample size in each category. Daily use for ≥3 hours was defined as high use . As very few participants reported long duration of phone calls, we categorized responses as 0-5 minutes, 6-15 minutes, 16-59 minutes, and ≥1 hour average daily use.
Mediator—Insufficient Sleep on Weekdays and Weekends
Adolescents provided information on when they usually got into bed, how long it took them to fall asleep, and what time they usually woke up, separately for weekdays and weekends. Sleep duration was derived from these responses. A sleep duration of less than 9 hours per day was defined as insufficient sleep, based on the recommendation of the US National Sleep Foundation for school-aged children .
Outcome—BMI z Score and Overweight Status at About 12 Years
We measured height (m) and weight (kg) using Tanita BC-418 Body Composition Analyzer. We used age- (in days) and sex-specific BMI z scores based on the 2007 World Health Organization growth reference by interpolating the World Health Organization references on a daily scale . Participants with implausible BMI z scores (SD ≥4 away from the mean) were excluded (n=1). We also defined adolescent overweight and obesity as a BMI for age and sex corresponding to an adult BMI ≥25 and 30 kg/m2, respectively, using the International Obesity Task Force cutoffs [ ]. Owing to the relatively small number of adolescents with obesity (79/1473, 5.36%), we combined obese and overweight into one category of overweight (total n=377).
Demographic information including age, sex, ethnicity (combined into White, Black, Asian, Mixed, and Other), and parental education and occupation was captured in the SCAMP assessment. Parental education was categorized in a binary form as follows: at least one parent with higher education and no parent with higher education. We used the Office for National Statistics classification of parental occupation, categorizing it into 3 levels . Each child was allocated the higher level of either parent. Dietary factors were also considered as potential confounders. The participants were asked whether they normally eat breakfast. Responses were categorized as yes or no.
Ordinal logistic regression yielded adjusted odds ratios (ORs) and 95% CIs for each exposure variable (phone calls, internet use on mobile phones, and video gaming) in relation to demographic characteristics and school type (state or independent). Multilevel linear regression was used to assess the associations between each of the exposure variables and BMI z score, adjusted for potential confounders and school clustering effect. Multilevel logistic regression was used to assess the associations between each exposure variable and overweight.
Potential confounders included age, sex, ethnicity, parental education, parental occupation, and breakfast eating, selected based on directed acyclic graphs (DAGs) . DAG is a simple and transparent way to identify and demonstrate causal relationships between variables via graphs. Confounders were defined as the mutual causes of exposure and outcome variables. Model 1 showed unadjusted associations between each of the exposure variables and BMI z score or overweight. Model 2 was adjusted for age, sex, and ethnicity. In addition, we adjusted for socioeconomic status (SES) indicators (parental education and parental occupation) and breakfast eating in model 3. Missing values for covariates were assigned to a separate missing category for each covariate but were not excluded from the analysis. We also examined whether these associations were modified by sex, ethnicity, or parental occupation based on the significance of the interaction terms. We also assessed the associations between weekday and weekend digital technology use and BMI z scores. The z test for equality of regression coefficients was performed to assess whether associations with weekday use and weekend use differed.
Causal mediation analysis was performed to examine the role of insufficient sleep as a potential mediator between digital technology use and BMI z scores. Mediation effects of insufficient sleep on weekdays and weekends were analyzed separately. We used VanderWeele formula to compute the direct effect (ie, the effect of digital technology use on BMI z score independent of insufficient sleep), indirect effect (ie, the effect of digital technology use on BMI z score via insufficient sleep), total effect, and the proportion mediated by insufficient sleep . SEs for mediation analysis were estimated using a bootstrap procedure with 5000 bootstrap replications to obtain a 95% CI. All analyses were performed using STATA version IC/13.1 for Windows (Stata Corp).
The North West Haydock Research Ethics Committee approved the SCAMP study protocol and subsequent amendments (reference number 14/NW/0347). Head teachers of schools consented to participation in SCAMP. Parents and adolescents were provided with written information about the study in advance and were given the opportunity to opt out of the research at any time. The study was conducted in accordance with the Declaration of Helsinki.
The median age of our study sample was 12.06 (IQR 11.79-12.33) years. The sample was diverse in terms of ethnicity and SES (). Almost a fifth of the cohort (1146/6616, 17.32%) spent ≥3 hours per day on internet use and 11.08% (733/6616) on video gaming. A total of 6.71% (444/6616) spent ≥1 hour per day on phone calls. The correlation between measures (eg, phone calls) of digital technology use was low (Cohen <0.3). Of the original 6616 participants in SCAMP, 1473 (22.26%) had measured height and weight. The mean BMI z score in our sample was 0.43 (SD 1.21). We could not find a substantial difference between these participants and the entire cohort in terms of sociodemographic characteristics and digital technology use [ ].
|Sociodemographic characteristics||Overall (N=6616)||Bio-zone participants with measured height and weight (n=1473)|
|Age (years), median (IQR)||12.06 (11.79-12.33)||12.18 (11.94-12.44)|
|Sex, n (%)|
|Male||3147 (47.57)||747 (50.71)|
|Female||3469 (52.43)||726 (49.29)|
|Ethnicity, n (%)|
|White||2820 (42.62)||643 (43.65)|
|Black||1016 (15.36)||189 (12.83)|
|Asian||1758 (26.57)||425 (28.85)|
|Mixed||740 (11.19)||160 (10.86)|
|Other||62 (0.94)||17 (1.15)|
|Missing or not interpretable||220 (3.33)||39 (2.65)|
|Parental higher education, n (%)|
|At least one||3677 (55.58)||891 (60.49)|
|None||1200 (18.14)||265 (17.99)|
|Missing||1739 (26.28)||317 (21.52)|
|Parental occupation, n (%)|
|Managerial and professional||3426 (51.78)||867 (58.86)|
|Intermediate||1480 (22.37)||300 (20.37)|
|Routine and manual||1040 (15.72)||201 (13.65)|
|Missing||670 (10.13)||105 (7.13)|
|School type, n (%)|
|State||5141 (77.71)||1035 (70.26)|
|Independent||1475 (22.29)||438 (29.74)|
|Phone calls, n (%)|
|0-5 min||2236 (33.8)||559 (37.95)|
|6-15 min||1943 (29.37)||405 (27.49)|
|16-59 min||867 (13.1)||190 (12.9)|
|>1 h||444 (6.71)||81 (5.49)|
|Missing||1126 (17.02)||238 (16.16)|
|Internet use on mobile phones, n (%)|
|0-30 min||2016 (30.47)||459 (31.16)|
|31-59 min||858 (12.97)||183 (12.42)|
|1-2 h||1094 (16.54)||253 (17.18)|
|>3 h||1146 (17.32)||264 (17.92)|
|Missinga||1502 (22.7)||314 (21.32)|
|Video gaming on any device, n (%)|
|0-30 min||3440 (51.99)||788 (53.49)|
|31-59 min||981 (14.83)||233 (15.82)|
|1-2 h||1252 (18.92)||286 (19.42)|
|>3 h||733 (11.08)||124 (8.42)|
|Missing||210 (3.17)||42 (2.85)|
|BMI z score, mean (SD)||N/Ab||0.43 (1.21)|
aParticipants who did not own a mobile phone were recorded as missing data.
bN/A: not applicable.
Associations Between Digital Technology Use and Other Variables
shows that after mutual adjustment of all sociodemographic factors, participants at older ages reported longer duration of phone calls and internet use on mobile phones. Girls reported longer duration of phone calls but shorter duration of video gaming compared with boys. Compared with participants of White ethnicity, participants of Black ethnicity reported longer call duration and internet use on mobile phones, but the duration of video gaming was similar between White and Black participants. Asian participants reported lower levels of digital technology use. Participants whose parental occupation was classified as routine and manual reported longer duration of phone calls and video gaming than participants with parents in managerial and professional occupation. Participants from state schools reported longer duration of digital technology use than those from independent schools.
shows that participants who spent ≥1 hour per day on mobile phone calls had 0.42 (95% CI 0.13-0.71) higher BMI z scores than those reporting <5 minutes of use, after adjusting for age, sex, ethnicity, parental education, parental occupation, and breakfast eating (model 3). and show that participants who used the internet on mobile phones for more than 3 hours per day had 0.30 (95% CI 0.11- 0.48) higher BMI z scores and 60% ([1.6-1]/1) higher odds of being overweight (OR 1.6, 95% CI 1.09-2.34) than those reporting <30 minutes of use, respectively. Video gaming for ≥3 hours per day was also associated with higher BMI z scores (β=.26, 95% CI 0.03-0.5) and greater odds of being overweight (OR 1.62, 95% CI 1.03-2.53). The associations between digital technology use and BMI z score and being overweight were not modified by sex, ethnicity, or parental occupation (P values for interaction ranged from .08 to .89).
shows that internet use on mobile phones and video gaming at high level (ie, more than 3 hours per day) on weekdays were associated with higher BMI z score (β=.30, 95% CI 0-0.61; β=.35, 95% CI 0.07-0.63, respectively). However, associations were not evident between high weekend use and BMI z score. The association between video gaming on weekdays and BMI z score was stronger than that between weekend use and BMI z score (P=.009). Stronger association of weekday internet use on mobile phones than weekend use with BMI z score was also observed, although the difference was not significant (P=.07).
shows that insufficient sleep (both on weekdays and weekends) partly mediated (proportion of mediation ranged from 8.6% to 17.8%) the association between high digital technology use and BMI z score by an indirect effect. High use of digital technology remained associated with BMI z score after adjusting for insufficient sleep, shown as the direct effect.
|Characteristics||Phone callsb, aORc (95% CI)||Internet use on mobile phonesd, aOR (95% CI)||Video gaming on any devicee, aOR (95% CI)|
|Per year increase||1.16 (1.02-1.32)||1.67 (1.46-1.90)||1.00 (0.88-1.14)|
|Male||1 (reference)||1 (reference)||1 (reference)|
|Female||1.51 (1.36-1.67)||0.97 (0.88-1.08)||0.17 (0.15-0.19)|
|White||1 (reference)||1 (reference)||1 (reference)|
|Black||1.70 (1.46-1.97)||1.64 (1.41-1.90)||1.10 (0.95-1.27)|
|Asian||0.65 (0.57-0.74)||0.77 (0.67-0.88)||0.63 (0.56-0.72)|
|Mixed||1.24 (1.05-1.46)||1.24 (1.05-1.47)||1.24 (1.06-1.46)|
|Other||1.64 (0.99-2.73)||1.08 (0.64-1.82)||0.89 (0.55-1.44)|
|Parental higher education|
|Yes||1 (reference)||1 (reference)||1 (reference)|
|No||0.99 (0.86-1.14)||1.12 (0.97-1.30)||1.04 (0.90-1.19)|
|Managerial and professional||1 (reference)||1 (reference)||1 (reference)|
|Intermediate||1.09 (0.95-1.24)||1.13 (0.98-1.29)||1.19 (1.04-1.36)|
|Routine and manual||1.19 (1.02-1.39)||1.13 (0.97-1.33)||1.27 (1.09-1.47)|
|Independent||1 (reference)||1 (reference)||1 (reference)|
|State||1.88 (1.65-2.15)||2.7 (2.35-3.11)||2.78 (2.40-3.22)|
aAssociations adjusted for all other independent variables in the table.
bCategorized as 0 to 5 minutes, 6 to 15 minutes, 16 to 59 minutes, and >1 hour.
caOR: adjusted odds ratio, indicating the odds of being in higher categories of each characteristic of digital technology use associated with the independent variables relative to the reference group.
dCategorized as 0 to 30 minutes, 31 to 59 minutes, 1 to 2 hours, and >3 hours.
eCategorized as 0 to 30 minutes, 31 to 59 minutes, 1 to 2 hours, and >3 hours.
|Digital technology use||Model 1b, β (95% CI)||Model 2c, β (95% CI)||Model 3d, β (95% CI)|
|0-5 min||0 (reference)||0 (reference)||0 (reference)|
|6-15 min||.05 (−0.11 to 0.20)||.06 (−0.10 to 0.21)||.06 (−0.10 to 0.21)|
|16-59 min||.04 (−0.16 to 0.24)||.05 (−0.15 to 0.25)||.06 (−0.14 to 0.26)|
|>1 h||.40 (0.12 to 0.69)||.42 (0.13 to 0.71)||.42 (0.13 to 0.71)|
|P value for trend||.03||.03||.03|
|Internet use on mobile phones|
|0-30 min||0 (reference)||0 (reference)||0 (reference)|
|31-59 min||.17 (−0.04 to 0.37)||.14 (−0.07 to 0.34)||.15 (−0.06 to 0.35)|
|1-2 h||.10 (−0.09 to 0.28)||.07 (−0.11 to 0.26)||.08 (−0.11 to 0.26)|
|>3 h||.31 (0.13 to 0.50)||.30 (0.11 to 0.49)||.30 (0.11 to 0.48)|
|P value for trend||.003||.005||.006|
|Video gaming on any device|
|0-30 min||0 (reference)||0 (reference)||0 (reference)|
|31-59 min||.27 (0.10 to 0.45)||.23 (0.05 to 0.41)||.22 (0.04 to 0.40)|
|1-2 h||.18 (0.02 to 0.35)||.10 (−0.07 to 0.28)||.10 (−0.07 to 0.28)|
|>3 h||.37 (0.14 to 0.60)||.27 (0.03 to 0.50)||.26 (0.03 to 0.50)|
|P value for trend||<.001||.02||.03|
aAge- and sex-specific BMI z score, using 2007 World Health Organization growth references for 5-19 years.
cAdjusted for age, sex, and ethnicity.
dAdditionally adjusted for parental education, parental occupation, and breakfast eating.
|Digital technology use||Model 1b, ORc (95% CI)||Model 2d, OR (95% CI)||Model 3e, OR (95% CI)|
|0-5 min||1 (reference)||1 (reference)||1 (reference)|
|6-15 min||1.03 (0.76-1.40)||1.03 (0.75-1.40)||1.02 (0.74-1.39)|
|16-59 min||1.18 (0.80-1.73)||1.2 (0.81-1.77)||1.23 (0.83-1.83)|
|>1 h||1.46 (0.87-2.45)||1.43 (0.84-2.43)||1.45 (0.85-2.48)|
|P value for trend||.15||.17||.15|
|Internet use on mobile phones|
|0-30 min||1 (reference)||1 (reference)||1 (reference)|
|31-59 min||1.34 (0.89-2.01)||1.33 (0.88-2.00)||1.35 (0.89-2.05)|
|1-2 h||1.22 (0.84-1.77)||1.22 (0.84-1.78)||1.26 (0.86-1.85)|
|>3 h||1.55 (1.08-2.24)||1.57 (1.08-2.28)||1.60 (1.09-2.34)|
|P value for trend||.03||.03||.02|
|Video gaming on any device|
|0-30 min||1 (reference)||1 (reference)||1 (reference)|
|31-59 min||1.42 (1.01-1.98)||1.38 (0.98-1.96)||1.38 (0.98-1.96)|
|1-2 h||1.35 (0.98-1.86)||1.29 (0.91-1.82)||1.29 (0.91-1.83)|
|>3 h||1.68 (1.11-2.56)||1.60 (1.03-2.49)||1.62 (1.03-2.53)|
|P value for trend||.005||.03||.03|
aOverweight (including obesity) was defined as a BMI for age and sex corresponding to an adult BMI of ≥25 kg/m2.
cOR: odds ratio.
dAdjusted for age, sex, and ethnicity.
eAdditionally adjusted for parental education, parental occupation, and breakfast eating.
|Digital technology use||Weekday use, β (95% CI)||Weekend use, β (95% CI)|
|0-5 min||0 (reference)||0 (reference)|
|6-15 min||−.09 (−.29 to .10)||.01 (−.18 to .19)|
|16-59 min||−.13 (−.39 to .13)||.28 (.05 to .51)|
|>1 h||.13 (−.30 to .56)||.25 (−.11 to .61)|
|P value for trend||.77||.04|
|Internet use on mobile phones|
|0-30 min||0 (reference)||0 (reference)|
|31-59 min||.16 (−.09 to .41)||0 (−.22 to .23)|
|1-2 h||.05 (−.23 to .33)||−.08 (−.33 to .18)|
|>3 h||.3 (.00 to .61)||−.02 (−.31 to .27)|
|P value for trend||.05||.73|
|Video gaming on any device|
|0-30 min||0 (reference)||0 (reference)|
|31-59 min||.29 (.08 to .49)||−.09 (−.28 to .10)|
|1-2 h||.17 (−.08 to .41)||−.10 (−.31 to .12)|
|>3 h||.35 (.07 to .63)||−.11 (−.35 to .14)|
|P value for trend||.01||.40|
aAdjusted for age, sex, ethnicity, parental education, parental occupation, breakfast eating, and school clustering effect; weekday and weekend use also mutually adjusted.
|Mediator and exposure||Indirect effect, β (95% CIb)||Direct effect, β (95% CI)||Total effect, β (95% CI)||Proportion mediated (%)|
|Insufficient sleep on weekdays|
|0-5 min||0 (reference)||0 (reference)||0 (reference)||N/Ac|
|6-15 min||.02 (−.01 to .02)||.03 (−.13 to .19)||.05 (−.13 to .19)||N/A|
|16-59 min||.03 (.00 to .06)||.10 (−.11 to .32)||.13 (−.08 to .34)||N/A|
|>1 h||.05 (.03 to .12)||.48 (.19 to .76)||.53 (.27 to .83)||8.6|
|Internet use on mobile phones|
|0-30||0 (reference)||0 (reference)||0 (reference)||N/A|
|31-59||.02 (.00 to .05)||.12 (−.09 to .33)||.14 (−.07 to .35)||N/A|
|1-2 h||.03 (.01 to .06)||.06 (−.13 to .25)||.09 (−.10 to .27)||N/A|
|>3 h||.05 (.01 to .10)||.28 (.08 to .48)||.33 (.14 to .53)||15.9|
|Video gaming on any device|
|0-30 min||0 (reference)||0 (reference)||0 (reference)||N/A|
|31-59 min||.01 (−.01 to .03)||.21 (.04 to .39)||.22 (.04 to .40)||N/A|
|1-2 h||.02 (.01 to .05)||.12 (−.05 to .31)||.14 (−.03 to .33)||N/A|
|>3 h||.05 (.02 to .09)||.26 (−.01 to .53)||.3 (.04 to .57)||15|
|Insufficient sleep on weekends|
|0-5 min||0 (reference)||0 (reference)||0 (reference)||N/A|
|6-15 min||.01 (−.01 to .02)||.04 (−.12 to .20)||.05 (−.12 to .20)||N/A|
|16-59 min||.03 (.00 to .05)||.1 (−.10 to .32)||.13 (−.08 to .34)||N/A|
|>1 h||.05 (.04 to .13)||.48 (.19 to .76)||.53 (.28 to .83)||9.2|
|Internet use on mobile phones|
|0-30 min||0 (reference)||0 (reference)||0 (reference)||N/A|
|31-59 min||.01 (−.01 to .02)||.12 (−.09 to .33)||.13 (−.09 to .32)||N/A|
|1-2 h||.03 (.00 to .06)||.06 (−.13 to .24)||.09 (−.11 to .27)||N/A|
|>3 h||.05 (.00 to .09)||.29 (.08 to .48)||.33 (.12 to .52)||14.3|
|Video gaming on any device|
|0-30 min||0 (reference)||0 (reference)||0 (reference)||N/A|
|31-59 min||.01 (−.01 to .03)||.22 (.04 to .39)||.22 (.05 to .41)||N/A|
|1-2 h||.02 (.01 to .07)||.13 (−.05 to .31)||.15 (−.02 to .34)||N/A|
|>3 h||.05 (.03 to .12)||.25 (−.03 to .52)||.31 (.04 to .58)||17.8|
aAdjusted for age, sex, ethnicity, parent education, parent occupation, breakfast eating, and school type.
bCI was obtained by 5000 bootstrap resamples.
cN/A: not applicable.
Our results showed that Black participants, those with lower SES in terms of parental occupation, and those from state schools use digital technology to a greater extent than White participants, those with higher SES, and those from independent schools, respectively. This study also found that high use of digital technology, including internet use on mobile phones and video gaming, was associated with higher BMI z scores and greater odds of being overweight in adolescents, using objective height and weight measurements. The associations were consistent across sex, ethnicity, and SES. The BMI z score was more strongly related to high weekday digital technology use than high weekend use. Insufficient sleep on weekdays and weekends partly mediated the associations between high digital technology use and BMI z scores.
In line with previous research [, ], this study found that adolescents whose parents had management and professional occupations used digital technology for a shorter duration. These parents may be more aware of the adverse health consequences of excessive digital engagement and potentially exert more control on and guidance to their children to avoid overuse. Moreover, they may be equipped with advanced digital skills to manage their children’s technology activities. Parents with lower SES, who can be under greater economic and time pressures, likely rate the restriction of children’s screen time and support of children’s web-based activities as relatively low priority compared with other life pressures [ ]. Compared with White adolescents, Black adolescents used mobile phones for a longer duration. Black adolescents may have a sense of being marginalized in a school with primarily White students (as in our sample) [ ]. The use of mobile phones to connect with family and friends is convenient to broaden and intensify their social support networks. This argument might also explain the different association patterns between ethnicity and video gaming, as the role of video gaming in social support is less pronounced than mobile phone use in general. Participants in state schools used digital technology more than those in independent schools, which is consistent with another study on British adolescents [ ]. Although the state schools in our study generally have more restrictions on digital technology use than independent schools (ie, students are strictly not allowed to use their mobile phones during school time), participants from state schools might use digital technology more during weekday evenings.
Screen time, such as television watching, mobile phone use, and video gaming, has been associated with adolescent obesity [, , ]. However, inferences from these studies are limited because of the bias of self-reported weight and height as well as a lack of differentiation between weekday and weekend use. To our knowledge, this is the first study to distinguish between weekday and weekend use and investigate a plausible mechanism (insufficient sleep) to help explain the association between digital technology use and adolescent obesity.
This study adds to the literature by showing that weekday use of digital technology might play a more important role than weekend use in individual differences in BMI z scores. Previous research has found that biomarkers related to obesity (eg, insulin and Homeostatic Model Assessment for Insulin Resistance) are more strongly associated with weekday digital technology use than weekend use in adolescents . This is possibly because of the metabolic risks attributed to prolonged sitting bouts already required during class time on weekdays [ ]. This study also adds to the literature by showing that insufficient sleep (both on weekdays and weekends) partly explained the associations between digital technology use and BMI z scores through an indirect effect. Plausible explanations include that excessive use of digital technology may displace sleep time, particularly on weekdays. This is because most schools restrict the use of mobile phones during school time, and participants need to wake up early to attend school; therefore, the main time during which participants can use their phones on weekdays is during the evenings and nights. Although direct sleep time displacement is less likely on weekend nights, sleep loss might also be because of psychological and physiological arousal by the content as well as melatonin suppression caused by screen light [ , ]. Notably, the association between digital technology use and BMI z score was still evident after adjusting for insufficient sleep (ie, direct effect), indicating that the underlying mechanisms may extend beyond sleep deprivation. High engagement in digital technology may also displace physical activity, resulting in obesity [ ]. Mobile phone and internet addiction, and video game dependency, may act as a manifestation of psychosocial stress, as digital technology may provide a means for adolescents to be distracted from stressful experiences [ - ]. The stress-induced hypothalamic-pituitary-adrenal axis could stimulate appetite and disrupt metabolism and thus promote weight gain [ , ].
This study has several strengths and weaknesses. It is based on the largest prospective cohort of adolescents worldwide, with detailed data on mobile phone technologies. The cohort is representative of school-aged children across Greater London. Analyses used objective measurements for height and weight and did not rely on self-reported information. The study was able to distinguish between weekday and weekend use and investigate a plausible mechanism of insufficient sleep to aid understanding of the association between digital technology use and adolescent obesity. However, this study is subject to residual confounding because of the nature of the observational study design, although confounders were rigorously selected on a scientific basis (ie, based on the DAG). In addition, information on physical activity, other dietary factors (eg, snacking), and psychosocial stressors was not available; therefore, other potential mechanisms could not be considered. We analyzed cross-sectional baseline data of the SCAMP sample; thus, we were not able to detect the temporal sequence of digital technology use and obesity. Longitudinal studies have shown that greater screen time is associated with increased changes in BMI and a higher risk of obesity in adolescents [, ]. However, we cannot exclude the possibility of obesity leading to sedentary behaviors, such as high engagement in digital technology [ ]. Baseline data were collected a few years ago and may, therefore, not reflect the contemporary digital environment, which is rapidly changing. However, we expect our findings to be relevant as screen time has been consistently associated with adolescent obesity in the literature [ , , , ]. The age range of our participants was small; therefore, our findings may not be generalizable to older adolescents. Finally, information on mobile phone use, video gaming, and sleep was derived from self-reports, possibly resulting in social desirability bias and measurement error.
This study suggests several avenues for future research. We plan to analyze longitudinal SCAMP data to investigate the temporal sequence of digital technology use and obesity and compare the relationship between digital technology use and obesity at different ages to create age-specific policy recommendations regarding digital technology activities. In addition, more comprehensive measurements of other obesity-related lifestyle factors such as physical activity and psychosocial stressors are important to unravel potential biological and behavioral mechanisms linking digital technology use and adolescent obesity. Objective measures of screen time and a secondary measure of sleep collected using a wearable fitness tracker (eg, Fitbit monitor), in addition to self-reported information, are needed to minimize self-report bias.
Our findings also have significant public health implications. The evidence of the potential impacts of digital technology overuse on adolescent obesity via insufficient sleep will allow the development of guidance for children and adolescents’ healthy use of digital technology to prevent obesity. It may be possible to maintain the level of digital technology use but mitigate its effects on health by altering patterns of use. In addition, the recent COVID-19 pandemic led to school closure, lockdown, and social distancing, which may substantially increase screen time and decrease opportunities for social and physical activity during this period . Schools, teachers, and parents should be vigilant of the secondary impacts of COVID-19 on risks of sleep disturbance and obesity and better advise their students and children on digital technology use.
This study shows the socioeconomic disparities of digital technology use among 11-to 12-year olds. High use of digital technology, that is, >3 hours per day, was associated with higher BMI z scores and greater odds of being overweight. In general, BMI z scores were more strongly related to high weekday digital technology use than high weekend use. The association between digital technology use and BMI z score was partly mediated by insufficient sleep, suggesting that the underlying mechanisms are multifactorial. Further research on longitudinal data is essential to explore the direction of relationships. Schools, parents, public health practitioners, policy makers, and adolescents themselves should be aware of the potentially adverse health effects of excessive digital technology engagement on sleep and obesity.
The SCAMP study is an independent research funded by the National Institute for Health Research (NIHR) Policy Research Program (PRP; Secondary School Cohort Study of Mobile Phone Use and Neurocognitive and Behavioural Outcomes/091/0212) via the Research Initiative on Health and Mobile Telecommunications—a partnership between public funders and the mobile phone industry. This study was supported by funds from the NIHR Health Protection Research Units in Environmental Exposures and Health and Chemical and Radiation Threats and Hazards (Health Protection Research Units -2012-10141), a partnership between Public Health England and Imperial College London, and the MRC Centre for Environment and Health (MR/L01341X/1). An extension to SCAMP was funded by the NIHR PRP. The funders of the study had no role in the design or conduct of the study or in the reporting of the SCAMP study results. The views expressed in this paper are those of the authors and not necessarily those of the NIHR, Public Health England, or Department of Health and Social Care. The authors would like to thank all the schools, parents, and pupils who participated in the SCAMP. In addition, they would like to thank Dr Danielle Ashworth, Rosemary Jenkins, Yasmin Bou Karim, Irene Chang, Margaret Douglass, Mark Ellis, Dr Daniela Fecht, Charlotte Fleming, Prof John Gulliver, Dr Nick Henriquez, Rosi Hirst, Stacey Jennings, Daphna Kesary, Dr Gemma Knowles, Dr Michael O Mireku, William Mueller, Aamirah Mussa, Julian Mutz, Dr Jonathan Pham, Dr Milagros Ruiz, Annalisa Sheehan, Dr Rachel B Smith, and Riitta Soininen from Imperial College London; Dr Benjamin Barratt, Dr Artemis Doutsi, Dr Rosamund Dove, and Dr Ian Mudway from King’s College London; Elizabeth Booth from Birkbeck, University of London; Dr John Rogers from Delosis Ltd, London; Mikaël J A Maes from University College London; Dr Marloes Eeftens from Swiss Tropical and Public Health Institute; and Dr Jürg Fröhlich and Marco Zahner from Swiss Federal Institute of Technology in Zürich for their contributions to the project.
CS conceived the analysis plan, cleaned and analyzed the data, interpreted the results, and wrote and revised the manuscript. ID, M Thomas, PE, MR, and M Toledano obtained funding for the study and revised the manuscript. M Toledano conceived the study, obtained funding, and inputted the data analysis and manuscript writing. All authors have approved the manuscript for publication.
Conflicts of Interest
Both M Toledano and PE received research funding in support of research on mobile phones and health. PE and M Toledano are principal investigators of the UK Cohort Study of Mobile Phone Use and Health, which was funded by the UK Department of Health and Social Care PRP (grant PR-ST-0713-00003) and was formerly jointly funded by industry and government, via the independent Mobile Telecommunications and Health Research Program.
- Children and parents: media use and attitudes report. Ofcom. 2017. URL: https://www.ofcom.org.uk/__data/assets/pdf_file/0020/108182/children-parents-media-use-attitudes-2017.pdf [accessed 2020-12-15]
- Mak K, Lai C, Watanabe H, Kim D, Bahar N, Ramos M, et al. Epidemiology of internet behaviors and addiction among adolescents in six Asian countries. Cyberpsychol Behav Soc Netw 2014 Nov;17(11):720-728. [CrossRef] [Medline]
- Lopez-Fernandez O, Honrubia-Serrano L, Freixa-Blanxart M, Gibson W. Prevalence of problematic mobile phone use in British adolescents. Cyberpsychol Behav Soc Netw 2014 Feb;17(2):91-98. [CrossRef] [Medline]
- Punamäki RL, Wallenius M, Nygård CH, Saarni L, Rimpelä A. Use of information and communication technology (ICT) and perceived health in adolescence: the role of sleeping habits and waking-time tiredness. J Adolesc 2007 Aug;30(4):569-585. [CrossRef] [Medline]
- Lemola S, Perkinson-Gloor N, Brand S, Dewald-Kaufmann JF, Grob A. Adolescents' electronic media use at night, sleep disturbance, and depressive symptoms in the smartphone age. J Youth Adolesc 2015 Feb;44(2):405-418. [CrossRef] [Medline]
- Seo DG, Park Y, Kim MK, Park J. Mobile phone dependency and its impacts on adolescents’ social and academic behaviors. Comput Hum Behav 2016 Oct;63:282-292. [CrossRef]
- Jackson LA, Zhao Y, Kolenic A, Fitzgerald HE, Harold R, Von Eye A. Race, gender, and information technology use: the new digital divide. Cyberpsychol Behav 2008 Aug;11(4):437-442. [CrossRef] [Medline]
- Söderqvist F, Hardell L, Carlberg M, Hansson Mild K. Ownership and use of wireless telephones: a population-based study of Swedish children aged 7-14 years. BMC Public Health 2007 Jun 11;7(1):105 [FREE Full text] [CrossRef] [Medline]
- Lau JT, Gross DL, Wu AM, Cheng K, Lau MM. Incidence and predictive factors of internet addiction among Chinese secondary school students in Hong Kong: a longitudinal study. Soc Psychiatry Psychiatr Epidemiol 2017 Jun 17;52(6):657-667. [CrossRef] [Medline]
- Sánchez-Martínez M, Otero A. Factors associated with cell phone use in adolescents in the community of Madrid (Spain). Cyberpsychol Behav 2009 Apr;12(2):131-137. [CrossRef] [Medline]
- Männikkö N, Ruotsalainen H, Miettunen J, Marttila-Tornio K, Kääriäinen M. Parental socioeconomic status, adolescents' screen time and sports participation through externalizing and internalizing characteristics. Heliyon 2020 Feb;6(2):e03415 [FREE Full text] [CrossRef] [Medline]
- Wang X, Xing W. Exploring the influence of parental involvement and socioeconomic status on teen digital citizenship: a path modeling approach. J Edu Technol Soc 2018;21(1):186-199 [FREE Full text]
- Lin Y, Liu Q. Perceived subjective social status and smartphone addiction tendency among Chinese adolescents: a sequential mediation model. Child Youth Serv Rev 2020 Sep;116:105222. [CrossRef]
- Devito M, Walker A, Birnholtz J, Ringland K, Macapagal K, Kraus A, et al. Social technologies for digital wellbeing among marginalized communities. In: Proceedings of the Conference Companion Publication of the 2019 on Computer Supported Cooperative Work and Social Computing. 2019 Presented at: CSCW '19: Computer Supported Cooperative Work and Social Computing; November 9-13, 2019; Austin TX USA p. 449-454. [CrossRef]
- La Torre G, Masala D, De Vito E, Langiano E, Capelli G, Ricciardi W, PHysical ActivitySocio-Economic Status collaborative group. Extra-curricular physical activity and socioeconomic status in Italian adolescents. BMC Public Health 2006 Jan 31;6(1):22 [FREE Full text] [CrossRef] [Medline]
- Kenney EL, Gortmaker SL. United States adolescents' television, computer, videogame, smartphone, and tablet use: associations with sugary drinks, sleep, physical activity, and obesity. J Pediatr 2017 Mar;182:144-149. [CrossRef] [Medline]
- Lajunen H, Keski-Rahkonen A, Pulkkinen L, Rose RJ, Rissanen A, Kaprio J. Are computer and cell phone use associated with body mass index and overweight? A population study among twin adolescents. BMC Public Health 2007 Feb 26;7(1):24 [FREE Full text] [CrossRef] [Medline]
- Arora T, Hussain S, Lam KH, Yao GL, Thomas GN, Taheri S. Exploring the complex pathways among specific types of technology, self-reported sleep duration and body mass index in UK adolescents. Int J Obes (Lond) 2013 Sep 8;37(9):1254-1260. [CrossRef] [Medline]
- Mitchell JA, Rodriguez D, Schmitz KH, Audrain-McGovern J. Greater screen time is associated with adolescent obesity: a longitudinal study of the BMI distribution from Ages 14 to 18. Obesity (Silver Spring) 2013 Mar;21(3):572-575 [FREE Full text] [CrossRef] [Medline]
- Thomée S, Härenstam A, Hagberg M. Mobile phone use and stress, sleep disturbances, and symptoms of depression among young adults--a prospective cohort study. BMC Public Health 2011 Jan 31;11(1):66 [FREE Full text] [CrossRef] [Medline]
- Hale L, Guan S. Screen time and sleep among school-aged children and adolescents: a systematic literature review. Sleep Med Rev 2015 Jun;21:50-58 [FREE Full text] [CrossRef] [Medline]
- Mireku MO, Barker MM, Mutz J, Dumontheil I, Thomas MS, Röösli M, et al. Night-time screen-based media device use and adolescents' sleep and health-related quality of life. Environ Int 2019 Mar;124:66-78 [FREE Full text] [CrossRef] [Medline]
- Chang A, Aeschbach D, Duffy JF, Czeisler CA. Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness. Proc Natl Acad Sci U S A 2015 Jan 27;112(4):1232-1237 [FREE Full text] [CrossRef] [Medline]
- Felső R, Lohner S, Hollódy K, Erhardt ?, Molnár D. Relationship between sleep duration and childhood obesity: systematic review including the potential underlying mechanisms. Nutr Metab Cardiovasc Dis 2017 Sep;27(9):751-761. [CrossRef] [Medline]
- Toledano M, Mutz J, Röösli M, Thomas M, Dumontheil I, Elliott P. Cohort Profile: The Study of Cognition, Adolescents and Mobile Phones (SCAMP). Int J Epidemiol 2019 Feb 01;48(1):25-26l [FREE Full text] [CrossRef] [Medline]
- Fritz MS, Mackinnon DP. Required sample size to detect the mediated effect. Psychol Sci 2007 Mar 04;18(3):233-239 [FREE Full text] [CrossRef] [Medline]
- Lane A, Harrison M, Murphy N. Screen time increases risk of overweight and obesity in active and inactive 9-year-old Irish children: a cross sectional analysis. J Phys Act Health 2014 Jul;11(5):985-991. [CrossRef] [Medline]
- Hirshkowitz M, Whiton K, Albert SM, Alessi C, Bruni O, DonCarlos L, et al. National Sleep Foundation's sleep time duration recommendations: methodology and results summary. Sleep Health 2015 Mar;1(1):40-43. [CrossRef] [Medline]
- de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ 2007 Sep 01;85(9):660-667 [FREE Full text] [CrossRef] [Medline]
- Cole T, Bellizzi M, Flegal K, Dietz W. Establishing a standard definition for child overweight and obesity worldwide: international survey. Br Med J 2000 May 06;320(7244):1240-1243 [FREE Full text] [CrossRef] [Medline]
- The National Statistics Socio-Economic Classification (NS-SEC). Office for National Statistics. 2010. URL: https://www.ons.gov.uk/methodology/classificationsandstandards/otherclassifications/thenationalstatisticssocioeconomicclassificationnssecrebasedonsoc2010 [accessed 2020-12-15]
- Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology 1999;10(1):37-48. [CrossRef]
- Valeri L, Vanderweele TJ. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: theoretical assumptions and implementation with SAS and SPSS macros. Psychol Methods 2013 Jun;18(2):137-150 [FREE Full text] [CrossRef] [Medline]
- Lampard AM, Jurkowski JM, Davison KK. The family context of low-income parents who restrict child screen time. Child Obes 2013 Oct;9(5):386-392 [FREE Full text] [CrossRef] [Medline]
- Junco R, Merson D, Salter DW. The effect of gender, ethnicity, and income on college students' use of communication technologies. Cyberpsychol Behav Soc Netw 2010 Dec;13(6):619-627. [CrossRef] [Medline]
- Jackson LA, von Eye A, Fitzgerald HE, Witt EA, Zhao Y. Internet use, videogame playing and cell phone use as predictors of children’s body mass index (BMI), body weight, academic performance, and social and overall self-esteem. Comput Hum Behav 2011 Jan;27(1):599-604. [CrossRef]
- Hardy LL, Denney-Wilson E, Thrift AP, Okely AD, Baur LA. Screen time and metabolic risk factors among adolescents. Arch Pediatr Adolesc Med 2010 Jul 05;164(7):643-649. [CrossRef] [Medline]
- Healy GN, Dunstan DW, Salmon J, Cerin E, Shaw JE, Zimmet PZ, et al. Breaks in sedentary time: beneficial associations with metabolic risk. Diabetes Care 2008 Apr 05;31(4):661-666. [CrossRef] [Medline]
- Rey-López JP, Vicente-Rodríguez G, Biosca M, Moreno LA. Sedentary behaviour and obesity development in children and adolescents. Nutr Metab Cardiovasc Dis 2008 Mar;18(3):242-251. [CrossRef] [Medline]
- Lam LT, Peng Z, Mai J, Jing J. Factors associated with internet addiction among adolescents. Cyberpsychol Behav 2009 Oct;12(5):551-555. [CrossRef] [Medline]
- Haug S, Castro RP, Kwon M, Filler A, Kowatsch T, Schaub MP. Smartphone use and smartphone addiction among young people in Switzerland. J Behav Addict 2015 Dec;4(4):299-307 [FREE Full text] [CrossRef] [Medline]
- Rehbein F, Kleimann M, Mößle T. Prevalence and risk factors of video game dependency in adolescence: results of a German nationwide survey. Cyberpsychol Behav Soc Netw 2010 Jan 14:269-277. [CrossRef]
- Hamer M, Steptoe A. Cortisol responses to mental stress and incident hypertension in healthy men and women. J Clin Endocrinol Metab 2012 Jan;97(1):29-34. [CrossRef] [Medline]
- Gundersen C, Mahatmya D, Garasky S, Lohman B. Linking psychosocial stressors and childhood obesity. Obes Rev 2011 May;12(5):54-63. [CrossRef] [Medline]
- Boone JE, Gordon-Larsen P, Adair LS, Popkin BM. Screen time and physical activity during adolescence: longitudinal effects on obesity in young adulthood. Int J Behav Nutr Phys Act 2007 Jun 08;4(1):26 [FREE Full text] [CrossRef] [Medline]
- Ekelund U, Brage S, Besson H, Sharp S, Wareham N. Time spent being sedentary and weight gain in healthy adults: reverse or bidirectional causality? Am J Clin Nutr 2008 Sep;88(3):612-617. [CrossRef] [Medline]
- Nightingale CM, Rudnicka AR, Donin AS, Sattar N, Cook DG, Whincup PH, et al. Screen time is associated with adiposity and insulin resistance in children. Arch Dis Child 2017 Jul 13;102(7):612-616 [FREE Full text] [CrossRef] [Medline]
- Laurson KR, Lee JA, Gentile DA, Walsh DA, Eisenmann JC. Concurrent associations between physical activity, screen time, and sleep duration with childhood obesity. ISRN Obes 2014 Mar 09;2014:204540-204546 [FREE Full text] [CrossRef] [Medline]
- Nagata JM, Magid HS, Gabriel K. Screen time for children and adolescents during the coronavirus disease 2019 pandemic. Obesity (Silver Spring) 2020 Sep 07;28(9):1582-1583 [FREE Full text] [CrossRef] [Medline]
|DAG: directed acyclic graph|
|NIHR: National Institute for Health Research|
|OR: odds ratio|
|PRP: Policy Research Program|
|SCAMP: Study of Cognition, Adolescents and Mobile Phones|
|SES: socioeconomic status|
Edited by R Kukafka; submitted 16.12.20; peer-reviewed by J Nagata, A Jolliff, S Smith; comments to author 08.02.21; revised version received 03.04.21; accepted 06.05.21; published 05.07.21Copyright
©Chen Shen, Iroise Dumontheil, Michael Thomas, Martin Röösli, Paul Elliott, Mireille Toledano. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 05.07.2021.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.