Physical Activity and Academic Achievement: An Umbrella Review

Background: This umbrella review aimed to summarise the evidence presented in systematic reviews and meta-analyses regarding the effect of physical activity on academic achievement of school-age children and adolescents. Methods: A comprehensive electronic search for relevant systematic reviews and meta-analyses were performed in Pubmed, Cochrane Library, Web of Science, Scopus, and Latin American and Caribbean of Health Sciences Information System, and reference lists of the included studies, from inception to May 2020. Studies were included if they were systematic reviews or meta-analyses, included school-age children or adolescents, the intervention included physical activity, and the outcome was the academic achievement. Two independent authors screened the text of potentially eligible studies and assessed the methodological quality of the studies using the AMSTAR 2 tool. Results: Forty-one systematic reviews and meta-analyses that examined the effects of physical activity on children and adolescents’ academic achievement were identified. Overall, the systematic reviews reported small positive or mixed associations between physical activity and academic achievement. From meta-analyses, it was observed that physical activity had null or small-to-medium positive effects on academic achievement. Chronic physical activity showed a medium positive effect on academic achievement, and acute physical activity did not demonstrate benefits. Conclusions: Physical activity seems not to be detrimental to school-age children and adolescents’ academic achievement, and may, in fact, be beneficial.


Introduction
It is well documented that physical activity (PA) has beneficial effects for both physical and mental health [1]. If sustainable development goals are to be met, it is essential that PA levels increase among all age groups but particularly among children and adolescents. However, physical inactivity is growing significantly for young people, in part due to the rise in school-related sedentary behaviours [2].
The World Health Organisation (WHO) recommends children and adolescents aged 5-17 years to achieve a minimum of 60 min of moderate-to-vigorous PA (MVPA) per day [3]. The accomplishment of recommended levels of PA is crucial for the development of cognitive, motor, and social skills, as well as good musculoskeletal health [4]. However, in 2016, it was estimated that only 34% of young Europeans met these recommendations [4]. At school, a study found that European pre-adolescent children (aged 10-12 years) spent an average of 65% of their school time in sedentary activities, while they spent Due to the significant implications for educational practices at the population level, a substantial body of research has been dedicated to understanding the effect that PA can have on students' cognition, classroom behaviour, and AA [22]. However, results from previous reviews have been inconsistent, which may be due to the variety of study designs employed in this area. For example, some reviews have used moderators, such as SES, family support, age, sex, psychological variables, nutritional status, while others have not. There have also been discrepancies in the measurement of AA with both standardised and non-standardised tests being used, as well as ambiguity in the definition of AA. Finally, a range of PA interventions has been included in previous reviews, such as extra-curricular PA, PE, active classrooms, active commuting to and from school, specific modalities, and acute or chronic PA. For these reasons, definitive conclusions cannot yet be drawn [17,23,24].
Therefore, this review aimed to summarise the available evidence presented in systematic reviews and meta-analyses regarding the effect of PA on AA of school-age children and adolescents, and explore the effect of PA programmes or modalities on AA in specific subjects.
The relevance of this umbrella review is to address the need for evidence to inform the development of future recommendations, strategies, and policies at different levels, particularly within the education sector.

Search Strategy
A comprehensive search for relevant systematic reviews was conducted in the following electronic databases: Pubmed, Cochrane Library, Web of Science, Scopus, and Latin American and Caribbean of Health Sciences Information System (LILACS), until May 2020. The full search strategy is described in Appendix A. For each selected database, we used the following search terms: ("physical activity" OR exercise OR "physical education" OR "active transport" OR "active mobility" OR walking OR cycling OR running OR training OR sport) AND ((academic OR school OR cognitive OR cognition) AND (achievement OR performance OR attainment OR function OR result)) AND (child* OR school-age OR schoolchildren OR adolescents OR youth), AND ("systematic review" OR meta-analysis). Specific search terms were modified according to the database requirements. There were no limits applied to the search and the reference lists of the included reviews were searched. The search was conducted by two authors, from inception until May 2020.

Study Selection
Two authors independently reviewed the search results and screened publications retrieved from databases and reference lists, according to predefined steps. First, articles were screened by the information from the title and abstract. Second, articles with potential relevance were retrieved for full-text review, and their eligibility for inclusion in the review was determined. Disagreements were resolved through discussion until consensus.
All studies that fulfilled the following eligibility criteria were included: type of studies-systematic reviews or meta-analysis; type of participants-school-age children or adolescents, i.e., six to 18 years old (studies that include data on younger or older students were not excluded if data could be interpreted for the eligible age range); type of interventions-any form of acute or chronic physical activity practice and; type of outcome-academic achievement.
Studies were excluded according to study type (editorials, comments, case reports, guidelines, conference abstracts, other reviews); studies which focused exclusively on participants with cognitive disabilities (e.g., autism, attention deficit hyperactivity disorder); studies without any form of PA; studies without measurement and quantification of the outcomes of interest; and studies without available full-text.
For this review, AA refers specifically to grades and standardised test results, and only reviews that reported this outcome were included. "Academic performance" (AP) is sometimes used as a synonym for AA. At the same time, it describes these outcomes plus other measures, such as attendance, classroom behaviour, time on task, or executive function. Cognitive outcomes are often included in reviews in this field, considering they play a role in academic success. We did not exclude studies that used "academic performance" or "cognitive function", where authors defined the term clearly and reported the specific outcomes included in our definition of AA. The setting for most of the included reviews was educational institutions or the community.

Data Extraction
Each selected review was independently evaluated by two authors to extract information regarding the study design, objectives, participants (type and age), number and type of included studies, PA intervention and outcome measures, the setting of intervention, subject-specific effects, the overall effect of PA on AA, and the effect sizes from the meta-analyses. If there were discrepancies in data extraction, the authors discussed until there was consensus.

Methodological Quality
Two independent investigators evaluated the methodological quality of each eligible systematic review, using the AMSTAR 2 tool [25]. This instrument has 16 items and enables appraisal of systematic reviews of randomised and non-randomised studies of healthcare interventions. Each study is rated according to critical domains that can affect the validity and the conclusion of the review. The critical domains considered for this review were protocol registration before the commencement of the review (item 2); adequacy of the literature search (item 4); justification for excluding individual studies (item 7); risk of bias from individual studies being included in the review (item 9); appropriateness of meta-analytical methods (item 11); consideration of the risk of bias when interpreting the results of the review (item 13); and assessment of the presence and likely impact of publication bias (item 15).
The studies were rated as 'high-quality' if no or only one non-critical weakness was present; moderate-quality´if there was more than one non-critical weakness;´low-quality´if there was only one critical flaw with or without non-critical weaknesses; and´critically low-quality' if there was more than one critical flaw with or without non-critical weaknesses. Any disagreements in the classification were resolved by discussion.

Study Selection
A total of 2314 references were identified in the initial search in electronic databases, and an additional ten references were identified through reference lists. After the duplicated studies were removed (n = 751), 1573 studies remained. After screening for the title and abstract, 1442 papers were excluded, and 131 studies were eligible for full-text reading, from which 90 were removed. Thus, we included 41 studies for qualitative synthesis (Figure 1).
Studies reported distinct definitions and instruments to assess AA, such as school grades, grade point average, standardised test scores, and subject-specific test scores (e.g., mathematics, reading, spelling).
Regarding the effects of PA on specific subjects, the included reviews evaluated reading, language, science, spelling, geography and mathematics. Mathematics was the subject where positive associations were found with more consistency [27,31,33,36,39,52,61].

Overall Academic Achievement Effect Subject-Specific Effect
Sibley and Etnier, 2003 [39] There is a small significant relationship between PA and AA in children.
Mathematics and verbal tests showed very small effects.

Strong et al., 2005 [42]
The addition of PE to the curriculum results in small positive gains in AP. The quasi-experimental data also suggest that allocating more curricular time to programs of PA does not negatively affect AA, even when the time allocated to other subjects is reduced. Some results also suggest a relative increase in AP per unit of time. Cross-sectional observations show a positive association between PA and AP.

Murray et al., 2007 [43]
Strong evidence suggests a lack of negative effects of PE programs on AA.
A positive trend toward increased arithmetic scores, but no significant changes in reading scores when compared with controls. No significant difference was noted between groups on the mathematics or composite basic battery scores.
Trudeau and Shephard, 2008 [44] Studies suggest that sport is more likely to benefit AA if offered in school rather than in other sports contexts. Adding time to "academic" or "curricular" subjects by taking time from PE programmes does not enhance grades and may be detrimental to health.

N/A
Fedewa and Ahn, 2011 [27] There was a small effect between PA and AA. Small to medium effects on mathematics, reading and language; null effect for science.

Rasberry et al., 2011 [9]
Mixed findings: PA is positively associated with AP (50.5% of the associations summarised) or has a null effect (48% of the associations).
Reported in subgroups according to the type of PA, however association or otherwise with AA not reported separately.

Singh et al., 2012 [45]
Authors found strong evidence of a significant positive relationship between PA and AP.
Positive effects on language skills, reading skills and a basic test battery; there was no significant difference for mathematics.

Haapala, 2012 [40] N/A
In four studies, three reported a positive effect of physical training on mathematical, reading, and language skills.

Lees and Hopkins, 2013 [41]
Aerobic PA is positively associated with AA. Non-significant effects for mathematics, language, and sciences.

Overall Academic Achievement Effect Subject-Specific Effect
Pucher, Boot and de Vries, 2013 [46] School health promotion intervention showed that interventions targeting PA and nutrition had a small to large effects on AP, and no negative effects occurred.
Effects of different kinds of interventions varied in size and across subjects like mathematics and language. Also, positive effects were observed in subjects from which time was taken for the intervention, e.g., for additional PE.
Busch et al., 2014 [47] In general, team sports had an overall positive impact on AA, while individual sports had no effect. N/A

Martin et al., 2014 [28]
Null effects were found for overall AA. Overall, null effects were found for mathematics, reading and language.
Conde and Tercedor, 2015 [26] Regarding the association between PA and AP, when it was assessed with Grade Point Average: 11 studies found a positive association, one study with no association and one study with a negative association; when evaluated with other instruments-four studies found at least one positive outcome.

N/A
Esteban-Cornejo et al., 2015 [48] Four studies showed no effect of PA on AP, one showed a negative association, and 11 showed a positive association. PA was more strongly associated with AA among girls than boys. In the two studies assessing PA and AA, there was a significant improvement.
One study showed significant improvement across all subjects, the other only in social sciences. A significant small-to-medium mean effect size of 0.367 was found for PA interventions on AA in adolescents. Larger effects were found for grades compared to standardised achievement tests. AP demonstrated a significant improvement with exercise in one of two studies.

N/A
One study showed a significant beneficial effect of PA on mathematics, while another study had a small effect without statistical significance.

Marques et al., 2017 [53]
Seven articles found a positive association between PE or school-based PA and AP, four found no association, and in one there was a positive association for third-grade students and a negative association for second-grade students. Objectively measured PA was inconsistently related to AA, although four studies of strong quality provided partial support for a positive relationship. Self-reported PA was positively associated with AA.
Mixed results between studies.
Daly-Smith et al., 2018 [58] The study of physically active learning found no effect on AA, while the study of classroom movement break found a positive effect on AA (when breaks were at least 10 min).
Mixed results for mathematics. Most of the studies that addressed AP using the scores obtained in non-standardised tests showed PA results improved AP. Interventions with higher exercise intensity and duration led to more pronounced improvements in AP.
Two studies found that active learning and the introduction of active breaks improved AP in children with lower grades. Three studies found that the benefits of their intervention programmes were more closely associated with improvements in mathematics, with particular emphasis on arithmetic, while not finding any relationship with reading comprehension.

Overall Academic Achievement Effect Subject-Specific Effect
Dudley and Burden, 2020 [37] Increasing the proportion of the curriculum allocated to PE had small positive effects on student learning.

Overall Academic Achievement Reported in the Meta-Analyses
Thirteen studies performed meta-analyses and provided effect sizes for AA, which are summarised in Table 3. In brief, the overall effect of different modes of PA had null [28,32,33] or small to medium [27,30,33,35,37] effects on AA.
Considering the type of PA on overall AA, increasing the allocation of PE at school had a small [37] or medium [27,39] effects; active classrooms, compared with traditional sedentary classrooms, had null [32] or medium [35] effects. Chronic PA had a medium effect on AA; however, acute PA had a null effect on AA [33].
Looking at the type of PA intervention on subject-specific achievement, active commuting to and from school was not associated with mathematics [23,24] and language [23]; increasing the allocation of PE at school had a small [39] or medium [27] effect on mathematics and reading; active classrooms, compared with traditional sedentary classrooms, had a null [35] or medium effect in mathematics [31], a null [35] or small [31] effect on reading, a null [31] or large [35] effect on language, a null effect on spelling and science [35], and a large effect on geography [35]. Table A1 shows the assessment of methodological quality for each study. Two studies were rated as 'high-quality', five studies with 'moderate-quality', 11 studies with 'low-quality', and 23 studies with 'critically low-quality'. The critical domains where studies did not meet the quality requirements were the registration of the protocol before the commencement of the review (item 2, n = 30), the consideration of the risk of bias when interpreting the results of the review (item 13, n = 22), and the assessment of the risk of bias from individual studies being included in the review (item 9, n = 15). In non-critical domains, the majority of reviews (item 10, n = 39) did not report the sources of funding of individual studies (item 10), did not report the complete Population, Intervention, Control group, and Outcome (PICO) components (item 1, n = 32), and did not explain the selection of study designs for inclusion (item 3, n = 28). There was total agreement between the reviewers on the methodological quality assigned to each study.

Main Results
This review summarises the evidence of 41 systematic reviews and meta-analyses examining the relationship between physical activity and AA in school-aged children and adolescents. Overall, the findings suggest that PA has a null or small to medium effect on AA in school-age children and adolescents.
Our results are consistent with the latest studies in this field [35]. However, the findings are mixed when reviews include experimental or longitudinal studies. Mathematics was the subject measured more frequently, compared with other subjects, with null or small to medium effects. It has been hypothesised that PA improves executive function, which, in turn, has an impact on inhibition, working memory, and cognitive flexibility, components associated to this subject [36]. Furthermore, improving cognitive skills, such as visuospatial skills, rapid automatised naming, and memory can contribute to arithmetic learning [63,64]. Despite these considerations, more studies are needed to assess the effect of long-term PA intervention on mathematics' performance [62], and also whether PA interventions that increase student's enjoyment of classes can promote psychological wellbeing-another factor in academic performance [36].
The allocation of time for PE at school has a small to medium effects, and active classrooms, compared with traditional sedentary classrooms, had mixed effects. These mixed findings could, in part, be explained by methodological issues, such as the variety of type and length of PA interventions, the heterogeneity of populations included, as well as the specific definition of AA and the method of measurement utilised. In addition, a lack of moderators, including SES, family support, age, sex, psychological variables, nutritional status, may have also contributed to these diverse findings [17].
The studies which assessed chronic PA also reported small positive effects after the observational period, which is a very promising finding. To investigate this further, it is important to conduct high-quality RCTs over a longer period of time, as well as long-term and large-sample size longitudinal observational studies.
Active commuting to and from school was not associated with improvements in mathematics [23,24] and language [23]. This may be explained by specific methodological factors, such as the different definitions employed across the primary studies and the various ways of measuring active commuting to and from school, including objective and subjective measures, the low frequency of participants who engage in active commuting to and from school, and the lack of moderators' assessment in the studies [23]. Other environmental factors may affect the low frequency of active commuting to and from school, such as air pollution, the existence of safe sidewalks, cycle paths, and routes to school, as well as perceived neighbourhood safety [24].
The majority of the included reviews were scored as 'low-quality' or 'critically low-quality' when assessing the methodological quality. For a study being rated as 'low' quality, it must present one critical weakness. Nevertheless, part of these studies failed to contain an explicit statement that the review methods were established before the conduct of the review, which does not necessarily mean the study has low-quality. Furthermore, AMSTAR 2 does not intend to provide a score, so these findings must be interpreted with caution.

Limitations
There are several limitations that must be considered when interpreting these findings. First, the included reviews were heterogeneous in the type of interventions (single or multiple interventions), the population included (children, adolescents, or both), type of study designs, outcome measures, and lack standardised definitions for outcomes, with the terms "academic achievement" and "academic performance" not clearly defined and sometimes used interchangeably [65], which may limit clear interpretations of the results.
Second, many reviews did not include risk of bias scoring, and did not report details of participant and assessor blinding, or provided insufficient information regarding concealment of allocation to the intervention or the control group [52] and, therefore, incorporated low-quality evidence to draw their conclusions, especially the oldest publications.

Implications for Practice
Innovative strategies are needed to provide adequate PA for children and adolescents. It would be beneficial to ascertain whether some types of PA, such as active breaks, could be recommended. Several European cities report less cycling and walking in the commute to and from school and policy actions to ensure that young people can actively travel to and from school, which could increase both PA and AA. Our findings may be useful to support national and local governments to design intersectoral approaches, involving the health, sports, and education sectors, that aim to improve both academic and physical development through the promotion of PA to students in and out of the school settings [4].
At the school level, teachers may need to be supported to apply innovative approaches and strategies to increase PA levels; teachers and school administrators could receive training on the well-understood beneficial effects of PA on health, and the likely beneficial effects on AA. This should include practical guidance on how to implement increased PA in schools, including how to involve students and parents in planning activities, with support from suitably qualified PE professionals to maximise its potentially beneficial effect on AA. Schools also need to have appropriate open spaces, materials and resources to provide diverse opportunities for regular PA for children and young people of all ages. Partnering with sports and community organisations can support the development of extra-curricular opportunities for physical activity [4].
At the community level, safe environments are needed to enable regular PA and active commuting to and from school. The availability of sports and fitness clubs/gyms, community youth clubs, such as scouts, can also increase opportunities while a range of age-and gender-specific forms of PA need to be available, especially for adolescents. Awareness of the availability of opportunities could be raised through information and communication technology, social media approaches, and community and youth organisations [4].

Future Research
There are a number of areas for potential future research that can be highlighted, with the most prominent being the need to establish the causality of the relationship between PA and AA. Insights may be gained from conducting high-quality RCTs with a number of different PA intervention arms in addition to a non-active control group [60].
Greater collaboration between exercise scientists and neuroscientists may also help to improve the quality of future research and attempts to understand more clearly the relationship between PA and AA [52].
Also, future research should incorporate standardised outcome measures and strive towards increased standardisation of PA interventions with attention to methodological rigour and consideration of relevant moderators.
While some reviews provide policy recommendations, these are often not sufficiently detailed or disseminated to the appropriate audiences [65]. Researchers may be advised to publish practical articles in association with education specialists so that their findings could actually be translated and implemented by school administrators.

Conclusions
PA seems not to be detrimental to school-age children and adolescents' AA, and may, in fact, be beneficial. Different types of PA appear to have different effects, with the most benefit gained from longitudinal programmes incorporating aerobic exercise. Policymakers at the national, local, and school level should be made aware of the latest evidence and encouraged to make changes accordingly. Funding: This work was support by a grant from the Government of the Russian Federation in the context of the WHO European Office for the Prevention and Control of Noncommunicable Diseases.

Conflicts of Interest:
The authors declare no conflict of interest.Disclaimer: J.B. and S.W. are staff members of the WHO. R.M. is a WHO consultant. The authors alone are responsible for the views expressed in this publication, and they do not necessarily represent the views, decisions, or policies of the institutions with which they are affiliated. Appendix A.2. Cochrane Database of Systematic Reviews "physical activity" OR exercise OR "physical education" OR "active transport" OR "active mobility" OR "active commuting" OR "active travel" OR walking OR cycling OR running OR training OR sport (Title Abstract Keyword) AND academic OR school (Title Abstract Keyword) AND achievement OR performance OR attainment OR function OR result OR cognitive OR cognition (Title Abstract Keyword) AND child* OR school-age OR schoolchildren OR adolescents OR youth (Title Abstract Keyword) AND "systematic review" OR meta-analysis (Title Abstract Keyword). Appendix A.3. Web of Science "physical activity" OR exercise OR "physical education" OR "active transport" OR "active mobility" OR "active commuting" OR "active travel" OR walking OR cycling OR running OR training OR sport [ ("physical activity" OR exercise OR "physical education" OR "active transport" OR "active mobility" OR "active commuting" OR "active travel" OR walking OR cycling OR running OR training OR sport) [