Digital Interventions for Universal Health Promotion in Children and Adolescents: A Systematic Review

The use of digital platforms presents unique opportunities for universal health promotion, such as physical activity, healthy nutrition and diets, and chronic disease prevention.^17–24  A variety of interventions are used to reach and engage SACA, including smartphone apps¹⁹ ; text-messaging for universal health promotion; social media and networking sites for peer-to-peer connectivity, counseling, and/or mental health and well-being support.^17,23,25,26  Platforms like video games and “exergames” or “active video games” have become increasingly popular to promote physical activity and reduce sedentary behavior.^21,27 

A previous scoping exercise conducted by the review authors highlighted the need to evaluate and compare different types of digital-based interventions. Previous reviews typically focused on one type of digital platform, such as text-messaging,¹⁸  preventing possible comparisons of effectiveness between different platforms. Most reviews also evaluated treatment, therapy or management-based interventions for existing or chronic conditions, which is not the focus of this review. If reviews did include interventions promoting healthy behaviors, intervention effects were not disaggregated from their treatment or therapy counterparts. This review aims to add to the growing knowledge base about digital-based interventions and to evaluate the use of digital platforms for universal health promotion in SACA, which include healthy living, physical activity, nutrition, reduction in screen use, sedentary behaviors, online risk exposure, and mental health globally.

This systematic review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses criteria and registered within the International Prospective Register of Systematic Reviews (CRD42020213361).²⁸  This review was initially designed to evaluate the effectiveness of both (1) nondigital interventions to reduce screen use and sedentary behavior, and (2) digital-based interventions for universal health promotion, in SACA. A single search strategy was used (Supplemental Information) and eligible studies were screened together until the abstraction stage, at which studies were abstracted and analyzed separately. Given the large number of studies, the review authors chose to report the evidence synthesis separately. For the synthesis on nondigital interventions for physical activity and screen reduction, please refer to the Supplemental Information.²⁹  As guidance, a socio-ecological conceptual framework for nondigital and digital interventions on health outcomes was proposed (Fig 1).

The search strategy (Supplemental Information) was developed according to population, intervention, comparator, and outcome criteria and conducted in PubMed, Embase, Medline, PsycINFO, The Cochrane Library, Cochrane Central Register of Controlled Trials, Cochrane Methodology Register, and the World Health Organization regional databases. Gray literature searches were also conducted. There were no limitations on geographical settings, publication language, or duration of intervention follow-up. The date of the final search was March 16, 2021. Additional details about search strategy development and other information sources are included in another supplement article.³⁰ 

A multistage screening process was conducted. Title-abstract screening was done by a single reviewer. Full-text review and data abstraction were conducted in duplicate. Further details about the screening and selection process can be found in a supplement article.³⁰  Studies were screened and selected according to specific eligibility criteria (Table 1). Eligible study designs included randomized controlled trials, quasi-experimental studies (QES), and nonrandomized trials that already assessed the feasibility of the intervention to evaluate the review question. Small pilot or feasibility trials without any follow-up trial were excluded.³¹ 

Interventions were defined as any planned action, program, or policy aiming to promote healthy lifestyle behaviors. Eligible comparisons were no intervention (placebo), standard arm of care (eg, existing school programs), or other intervention arms in the case of a multi-arm intervention. Studies comparing two types of digital platforms were excluded (ie, the comparator intervention is also implemented digitally). Interventions whose focus was treatment, therapy and/or management of existing chronic disease (eg, weight loss, or treatment of overweight and obesity, including mental health) were excluded. Interventions promoting sexual reproductive health (eg, human immunodeficiency virus prevention, and safe sex practices) and addressing substance use and addiction were not included as these topics were covered by other reviews in the current supplement.

Statistical analyses were conducted using Review Manager 5.4. Analyses for randomized controlled trials and QES were separated. A random effects model was used to mitigate heterogeneity. Meta-analyses were conducted for each outcome of interest, only where there was data from, at the minimum, 3 studies. Overall effect estimates were interpreted as statistically significant where the associated P value was <.05. Where multiple follow-up timepoints were reported, the timepoint immediately following completion of the intervention was used for analysis across all studies to minimize heterogeneity and increase generalizability. Estimates were not adjusted for clustering if cluster-randomized controlled trials reported adjusted estimates. Where appropriate, unit conversions were conducted. Sensitivity analyses were conducted for intervention type (computer-based, internet-based, mobile app, messaging platform, or active video games).

Risk of bias assessments were conducted according to Cochrane Effective Practice and Organization of Care guidelines for randomized trials, nonrandomized trials, controlled before-after and interrupted time series,³²  and the Cochrane Handbook for Systematic Reviews of Interventions.³¹  The authors independently assessed risk of bias for each study. These scores were compared, and a final decision was made.

The Cochrane Risk of Bias-2 tool was used to assess risk of bias in randomized controlled trials per the following domains: randomization process, deviations from the intended interventions (blinding of personnel, participants, and outcome assessment), missing outcome data, outcome measurement, the selection of the reported result, disclosure of funding, and conflicts of interest.^31,33  Studies were assigned an overall risk of bias judgement accordingly (low risk, high risk, some concerns).

Quasi-experimental study designs were assessed using the Risk of Bias Tool for Nonrandomized Studies of Interventions (ROBINS-I) tool.^31,34  Assessments were made per the following domains: bias due to confounding, bias in selection of study participants, bias in classification of interventions, bias due to deviations from intended interventions, bias due to missing data, bias in measurement of outcomes, bias in selection of the reported result. An overall judgement (low, moderate, serious, and critical risk) was assigned to each study.

A summary of the intervention effect and quality assessment for all pooled outcomes were produced using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach across 5 domains: study limitations, consistency of effect, imprecision, indirectness, and publication bias.³⁵  Evidence was downgraded from “high quality” by one level for serious (or by two levels for very serious) limitations, depending on assessments for risk of bias, indirectness of evidence, serious inconsistency, imprecision of effect estimates, or potential publication bias.

A database search produced 29 301 records; hand searching revealed another 168 records. Following removal of 9132 duplicates, 20 337 records were screened at the title-abstract stage, which identified 680 records for full-text review. Four hundred and seven records were excluded at the full-text review stage for reasons including wrong intervention type, wrong study design, wrong comparator, wrong patient population, or wrong outcomes (Supplemental Information). In total, 51 studies (146 articles) were included for nondigital interventions and 74 studies (127 articles) were included for digital-based interventions. This article will solely focus on digital-based interventions. For the synthesis on nondigital interventions for screen reduction, please refer to the other supplement.²⁹  Of the 74 studies evaluating digital interventions, 37 were included in meta-analysis. Where possible, the remaining studies were narratively synthesized. See Fig 2 for the study breakdown across exclusion reasons.

Of the 74 included studies, 46 were RCTs,^36–81  3 nonrandomized controlled trials^82–84  and 25^85–109  were QES (Table 2). Across all study designs, an overwhelming majority of studies (n = 61, 82%) were conducted in high-income countries: Australia (n = 8),^{37,40,61,62,66,81,83,84}  Belgium (n = 2),^57,58  Canada (n = 3),^43,52,78  France (n = 2),^69,100  England (n = 3),^54,82,86  Germany (n = 2),^74,80  Hong Kong (n = 3),^94,107,109  the Netherlands (n = 3),^49,51,72  New Zealand (n = 4),^48,75,79,93  Portugal (n = 2),^50,99  and the United States (n = 21),^{38,42, 44,46,53,55,56,63,64,67,70,85,87,89–92,96, 97,101,105}  and one each in Denmark,⁹⁵  Finland,³⁶  Italy,⁴¹  Japan,⁶⁸  South Korea,¹⁰²  Spain,¹⁰⁴  Norway.³⁹  One study was a multicountry study, comprised of only high-income countries (Austria, Belgium, Germany, Crete and Greece, and Sweden).⁸⁸  Thirteen studies were conducted in LMICs: Brazil (n = 5),^{47,60,65,76,77}  Thailand (n = 2),^71,73  China (n = 2),^59,108  Ghana,¹⁰⁶  Indonesia,¹⁰³  Malaysia,⁹⁸  and Turkey.⁴⁵ 

Most interventions were conducted in school settings (n = 46). Fourteen studies were conducted in the home environment,^{38,42,44,48,51,52, 54,63,72,73,81,87,97}  and 5 studies were conducted in a laboratory clinic setting.^{90,91,93,94,101}  The duration of most interventions ranged from a minimum of 2 weeks (daily frequency) to 20 months. Five studies conducted the intervention in a single day period, evaluating the use of exergames and active video games on physical activity measures compared with a rest state.^{90,91,93,94,101} 

The mean age of study participants ranged from 7.7 to 19.25 years. Twenty-nine studies recruited participants whose collective mean age or age range fell below 13 years of age.^{38,39,42,45,50–54,63,67,69,71,74, 76,80–82,90,91,93,94,97,102–104,107,109}  The remaining studies recruited participants in their teenage years, with the majority conducted with participants between the ages of 13 and 16 years. Of these, five recruited participants with a mean age between 17 and 19.9 years.^{55,65,68,73,79} 

Among the RCTs, 19 studies used an internet-based platform^{36,40,42–44,46, 47,49,51,55,63,65,66,68–71,73,74}  to conduct the intervention; 5 studies computer-based interventions,^{39,53, 57,58,64}  mostly CD-ROMs and computer programs; 8 studies used text messaging, SMS or other messaging platforms (eg, WhatsApp) to interact with the students and parents^{37,41,50,56,59,60,62,75} ; 5 studies developed or used a mobile app to deliver the interventions^{48,61,67,77,79} ; and 9 studies used active video games or exergames to reduce sedentary behavior and promote physical activity and healthy behaviors.^{38,45,52,54,72,76,78,80,81}  Of the QES studies, three studies used a computer-based intervention ^85,98,100 ; nine studies used an internet-based platform^{82–84,88,89,92,97,102,109} ; 10 studies used active video games or exergames^{86,90,91,93,94,101, 103–105,107} ; and two studies used messaging platforms^95,96  and three, mobile apps.^87,99,108  One study was multimedia and incorporated elements of internet-based, mobile, and messaging platforms.¹⁰⁶ 

Of the 46 RCTs, the majority of studies (n = 24) were rated with a risk of bias of some concerns.^{36,39–47,49,50,53,54,56–58, 65–68,71,77,80}  Seven studies were rated with a high risk of bias.^{38,51, 55,59,64,76,78}  The main concerns arose from selective reporting, incomplete outcome data, and high attrition. Fifteen studies were assessed with a low risk of bias.^{37,48, 52,60–63,69,70,72–75,79,81}  Random sequence generation was adequately reported by 29 of 46 studies; however, allocation concealment was unclear or not reported in most studies (n = 29). Only 17 studies reported methods of allocation concealment. By nature of the intervention design, authors of most studies reported that blinding of participants was not possible; only one study reported blinding of participants.⁶⁰  Blinding of personnel or outcome assessors was reported in 12 studies. Attrition bias (defined as a loss to follow up of >20%) was assessed as high risk in 19 studies; reasons provided by study authors include absence in school on day of post intervention assessment, refusal to be measured, left the school or program, and pregnancy.

Of the 28 QES and nonrandomized trials, all were assessed with a moderate risk of bias^{83–85, 87–92,94,96–101,103–106,108,109}  with the exception of 6 studies that were rated with a low risk of bias.^{82,86,93,95,102,107}  Main reasons for downgrading study quality to a moderate risk of bias include poor adjustment of confounding variables and bias due to deviations from intended interventions.

When compared with the control, digital-based interventions may result in a slight reduction of body fat percentage (MD −0.35%, 95% CI −0.63 to −0.06; participants = 615; studies = 5; I² = 0%; P = .02; low quality of evidence) (Fig 3). Of the five studies, two were online or web-based interventions,^44,53  two were active video games,^45,54  and one study used a specific messaging platform.⁶²  No major differences were indicated when analyzed by intervention type. The analysis for waist circumference measured in centimeters showed that the control group had a positive effect compared with the intervention (MD 0.35, 95% CI 0.08 to 0.62; participants = 3102; studies = 7; I² = 28%; P = .01; low quality of evidence) (Supplemental Information). Six of the 7 studies implemented online or web-based interventions on the phone or computer.^{47,49,60,61,69,71}  The interventions likely resulted in little to no difference in body mass index (BMI) as measured as kg/m² (MD −0.01, 95% CI −0.16 to 0.13; participants = 5285; studies = 14; I² = 33%; P = .88; moderate quality of evidence) (Supplemental Information) and may have resulted in little to no difference in BMI z-scores (MD −0.02, 95% CI −0.06 to 0.03; participants = 3306; studies = 11; I² = 95%; P = .47; low quality of evidence) (Supplemental Information). Sensitivity analyses were conducted to determine effects by intervention type (computer, internet-based, messaging platforms, mobile apps, and video games); no significant difference was noted across intervention types.

Eight QES reported anthropometric outcomes,^{82,85,86,96,100,102,107,108}  of which two studies^82,85  were excluded from analysis due to incomplete and missing data. Digital-based interventions likely resulted in little to no difference in BMI z-scores (MD −0.10, 95%CI −0.15 to −0.04; participants = 1398; studies = 3; I² = 0%; P = .0004; moderate quality of evidence) (Fig 4).

The evidence is very uncertain about effects on body fat percentage (MD −0.85, 95% CI: −1.33 to −0.37; participants = 1111; studies = 4; I² = 0%; P = .0006; very low quality of evidence) (Supplemental Information). Given the limited number of studies per anthropometric outcome, sensitivity analyses per intervention type were not possible. Other anthropometric outcomes can be found in the Supplemental Information.

Seventeen RCTs reported various outcomes on screen time and screen use,^{36,37,39,42,46,47,49,50,54,60–62,64, 65,67,72,74}  of which 5 were excluded^{46,47,64,65,72}  due to incomplete or missing data, no poolable outcomes, and different author-defined metrics of similar outcomes. When compared with control or no treatment groups, digital-based interventions probably result in little to no difference on screen time on all types of media, measured as minutes per day (MD −17.32, 95% CI −42.69 to 8.05; participants = 1894; studies = 8; I² = 90%; P = .18; moderate quality of evidence) (Fig 5). Sensitivity analysis by intervention type showed a possible improvement of mobile app interventions on screen time (MD −47.11, 95%CI −77.06 to −17.15; studies = 2), although the sample size is limited. Overall, intervention effects likely did not reduce screen time in terms of computer use and video gaming and TV and DVD use (Supplemental Information).

A total of eight RCTs^{36,42, 47,48,54,60,62,81} reported leisure-based sedentary behavior as defined by study authors. Of these, two studies were excluded from meta-analysis due to incomplete or missing data and no poolable outcomes.^36,47  Digital-based interventions, compared with the control or no treatment group, likely reduces sedentary behavior (mins per day) (MD −19.62, 95% CI −36.60 to −2.65; participants = 872; studies = 6; I² = 49%; P = .02; moderate quality of evidence) (Fig 6). Sensitivity analysis by intervention type showed that messaging-based interventions had a significant effect on sedentary behavior (MD −37.45; 95% CI −51.76 to −23.15; studies = 2), compared with internet-based platforms⁴²  or active video games.^54,81 

Twenty-four RCTs^{36–38,42,44,46,48,49,52–55,59,61–64,68, 70,72,73,78,80,81} reported physical activity outcomes. Eleven studies were excluded from the meta-analysis due to incomplete or missing data or no poolable outcomes.^{46,52,55,59,63,64, 68,70,72,78,80}  Overall, the evidence is very uncertain on the effect of digital-based interventions on moderate and vigorous physical activity (MVPA) (minutes per day) and daily step count (Supplemental Information). Sensitivity analysis indicated no significant effect for any particular type of intervention for both outcomes. Interventions had little to no effect on counts per minute, but the evidence is very uncertain (MD 51.45, 95% CI −17.08 to −117.98; participants = 1071; studies = 8; I² = 94%; P = .14; very low quality of evidence) (Supplemental Information). Sensitivity analysis showed that video games or exergames had a significant effect in the intervention group (MD 19.88, 95%CI 13.91 to 25.84; studies = 2), although the sample size is limited.

Twelve QES^{82,85,86,88–91,93,94,97,101,107}  reported physical activity outcomes. Of these, three studies were included in the analysis for MVPA (mins per day). The interventions likely increased MVPA among participants compared with control (MD 3.92, 95% CI 0.54 to 7.30; participants = 1210; studies = 3; I² = 1%; P = .02; low quality of evidence) (Fig 7).

Energy expenditure was another commonly reported outcome; however, high heterogeneity in used metrics and units restricted any pooled analysis. Overall, evidence suggests that energy expenditure increased during active video game or active computer play, compared with sedentary video games, leisure computer use, or rest.

Twenty-four RCTs^{38,39,41,43,44,46, 47,49–51,55–58,60–62,64,66,67,69,72,77,80}  reported outcomes for dietary intake, such as intake of macronutrients, total energy, and specific food groups. Of these, 12 studies were not included in the meta-analysis due to incomplete and missing data, and no poolable outcomes.^{39,43,46,47,51,55,61,64,66,72,77,80}  No significant effects were observed on fruit consumption measured as servings or pieces per day, daily energy intake, or fat intake (Supplemental Information). However, interventions did show a slight increase in the consumption of both fruit and vegetables, measured as servings or pieces per day (MD 0.63, 95% CI 0.21 to 1.04; participants = 1328; studies = 6; I² = 70%; P = .003; high quality of evidence) (Fig 8). Sensitivity analysis indicated that messaging-type interventions had a significant effect on fruit and vegetable intake (MD 0.98 servings per day, 95% CI 0.28 to 1.68, participants = 1093; studies = 3), compared with other media types.

Several RCTs also reported intake of snacks^47,49,51,72  and unhealthy food products, and sugar-sweetened beverages^{39,47,49,56,58,67,72} ; however, the majority of studies reported no significant effect of the intervention.

Fifteen studies assessed knowledge, intentions and perceptions of school-aged children and adolescents in nutrition, dietary intake and physical activity, with a few focused on sexual health and substance use education and prevention, targeting older adolescents.^{44,63, 66,76,77,80, 85,92,99,103–106,108,109}  Four RCT studies showed that overall knowledge acquisition in physical activity and nutrition recommenda tions and sexual health was greater in the intervention group, compared with the control group, as indicated by higher mean posttest scores.^44,63,66,80  Three of the 4 studies used a web-based platform^44,63,66  and one study used a gaming platform⁸⁰  to provide the education. No difference between intervention and control groups was noted for Chagas et al.⁷⁷  Similar findings were noted in the QES findings, whereby knowledge acquisition in nutrition, healthy lifestyles, and positive life perspectives was greatest in the intervention group compared with the control for the majority of studies.^{85,92,99,103–105,109} 

Eight studies examined mental health wellbeing, prevention of mental health illness, and promotion of mindfulness, self-esteem, and mood.^{40,75,76,79,83,84,109}  Two RCTs^40,75  and two QES studies^83,84  used interventions to prevent anxiety and depression in healthy SACA. One RCT⁴⁰  and both QES^83,84  implemented the MoodGYM intervention, an internet-based, self-paced interactive program to reduce the onset of anxiety and depressive symptoms. Calear et al⁴⁰  found the MoodGYM intervention condition reduced anxiety symptoms compared with the control group (P < .001). Another study also saw a significant rate of decline in depression scores in their female participants across the intervention period.⁸³  A similar study examined the same intervention with only male participants, and showed no significant change in scores from preintervention to immediate postintervention.⁸⁴  Whittaker et al,⁷⁵  also found no difference between intervention and control groups.

Among the included studies, there were no studies that assessed specific intervention effects on online risk exposure including, violent, hateful, and pornographic and/or unwanted sexual content; bullying, harassing or stalking, or sexual exploitation of SACA.

This review provides a comprehensive review of 74 studies across 25 countries, evaluating digital-based interventions for universal health promotion in healthy, normative BMI children and adolescents, and included 25 new studies conducted over the last 5 years.^{36,37,41,43,45,47,50,52,56,59–61,71, 72,75–80,103,105–109}  This review found that digital-based interventions resulted in a small but significant uptake in fruit and vegetable consumption, and a small but significant reduction in both body fat percentage and sedentary behavior. There is moderate to low certainty of evidence to suggest improvements in outcomes for physical activity and other dietary and anthropometric outcomes. As well, this review found that certain media types of intervention delivery had better effects on specific outcomes. Mobile apps showed improvements in screen time compared with other media platforms, while messaging platforms showed clear improvements in fruit and vegetable intake and sedentary behavior. Unsurprisingly, active video games showed improvements in counts per minute, compared with other media.

These findings support existing evidence of digital-based interventions for health promotion and behavior change.^20,110–112  Some reviews showed the effectiveness of active video games and exergames to encourage and promote physical activity and energy expenditure during leisure play time.^113–116  Others have noted the benefits of mobile phones to deliver short messages and engage with participants through apps and social media.^20,112  Mobile phones or internet-based apps are more portable, which may lead to greater potential in intervening and influencing behaviors regardless of setting.¹¹⁰  This portability can help to deliver interventions more effectively and consistently, leading to longer-lasting results. This review also noted that the type of digital platform varied depending on the age of participants. Interventions delivered through mobile phones, computers, and other internet platforms were largely implemented with older participants in their early to late teenage years. Comparatively, active video games were used among younger populations, typically 5 to 13 years of age. These observations suggest that older participants were selected due to the greater likelihood of possessing a mobile phone or having the capacity and knowledge to use phones, social media apps, or computer programs, thereby reducing possible confounders or the need for additional training. Active video and interactive games are considered more appropriate and fun for younger populations, possibly in an effort for greater attention, compliance, and adherence to the intervention.

Most interventions were multicomponent, involving parents or primary caregivers, especially when the mean participant age fell below the age of majority. Previous systematic reviews report that interventions may be more effective when caregivers were involved.^20,110  Caregiver involvement may be an important moderator, especially in early childhood during which they have less autonomy and control over their environments and are guided in their behaviors.¹¹⁰  However, all studies in this review did not isolate the effects of different intervention components. While this makes it difficult to evaluate the true effect of the digital component, this observation suggests that universal health promotion and behavioral change in school-aged children and adolescents may be most effective when designed as a multifaceted strategy. These may be considerations for future research when designing age-appropriate interventions.

It is important to note that there is growing evidence highlighting increased online risk, such as inappropriate content, unhealthy advertising, bullying, harassment and stalking, and sexual solicitation and exploitation, with increased use and interaction with online platforms.⁶  This review found no intervention study that specifically focused on preventing this online risk exposure. While cyberbullying is an acknowledged online risk, these studies were included in the mental health review of the same series.¹¹⁷  A recent rapid review showed that the majority of existing data assessing online risk exposure comes from survey and cross-sectional studies, and pointed to a substantial research gap in assessing the differences between online risk and harm, the long-term effects on development and well-being, and strategies and protective factors that can mitigate online risk exposure.⁶ 

To our knowledge, this is the first comprehensive systematic review to examine and compare digital media interventions through various mediums for universal health promotion in healthy SACA. However, this review unfortunately presents similar findings to an earlier scoping review conducted by the authors and related systematic reviews, pertaining to gaps in the evidence and methodologies. An overwhelming majority of studies are conducted in high-income settings; in this review, 82% of the interventions. Despite current knowledge that over the last decade, the number of people worldwide possessing a mobile phone and accessing the internet has increased significantly, there remains a stark “digital divide” between those living in LMICs and those in high-income countries.^6,118 

The heterogeneity of available data due to diverse interventions, a lack of standardization of outcome metrics, vague and diverse methodologies, and the use of subjective tools, such as self-reporting, also limit the findings of this review. Thus, generalizability of these findings proves difficult. Despite a large number of included studies, few were of high-quality, with the majority lacking description and/or implementation of robust methodologies. Consistent with existing literature,^20,112  risk of bias was notable across studies; the most frequent bias was failure to blind participants and personnel, attrition bias, and selective reporting. This limits and introduces a level of uncertainty regarding the effectiveness of these interventions. In view of these quality issues, the findings should be interpreted with caution.

Digital interventions offer great potential to promote universal health, and address health system shortcomings such as lack of access to services and resources, and education. This may have a significant impact on behaviors and lifestyles of our growing generations, lending to long-term, positive benefits in human health and development. Although this remains a relatively new research area with the majority of studies having been published in the last 5 to 10 years, largely in North America and Europe, continued efforts are needed to evaluate digital media as effective and accessible intervention platforms for health promotion in SACA globally. It is recommended that future research considers standardization of study design methodology and reporting of results and assesses the barriers and other factors that limit this type of research and program implementation in low- and middle-income populations and settings. Greater attention to interventions which assess the risk of online media including cyberbullying, sexual content, or exploitation and violence are needed to better support and inform universal health promotion efforts for SACA worldwide and inform policies and programs.

LMICs

low- and middle-income countries

SACA

school-aged children and adolescents