Impact of school policies on non-communicable disease risk factors – a systematic review

Background Globally, non-communicable diseases (NCDs) are identified as one of the leading causes of mortality. NCDs have several modifiable risk factors including unhealthy diet, physical inactivity, tobacco use and alcohol abuse. Schools provide ideal settings for health promotion, but the effectiveness of school policies in the reduction of risk factors for NCD is not clear. This study reviewed the literature on the impact of school policies on major NCD risk factors. Methods A systematic review was conducted to identify, collate and synthesize evidence on the effectiveness of school policies on reduction of NCD risk factors. A search strategy was developed to identify the relevant studies on effectiveness of NCD policies in schools for children between the age of 6 to 18 years in Ovid Medline, EMBASE, and Web of Science. Data extraction was conducted using pre-piloted forms. Studies included in the review were assessed for methodological quality using the Effective Public Health Practice Project (EPHPP) quality assessment tool. A narrative synthesis according to the types of outcomes was conducted to present the evidence on the effectiveness of school policies. Results Overall, 27 out of 2633 identified studies were included in the review. School policies were comparatively more effective in reducing unhealthy diet, tobacco use, physical inactivity and inflammatory biomarkers as opposed to anthropometric measures, overweight/obesity, and alcohol use. In total, for 103 outcomes independently evaluated within these studies, 48 outcomes (46%) had significant desirable changes when exposed to the school policies. Based on the quality assessment, 18 studies were categorized as weak, six as moderate and three as having strong methodological quality. Conclusion Mixed findings were observed concerning effectiveness of school policies in reducing NCD risk factors. The findings demonstrate that schools can be a good setting for initiating positive changes in reducing NCD risk factors, but more research is required with long-term follow up to study the sustainability of such changes. Electronic supplementary material The online version of this article (doi:10.1186/s12889-017-4201-3) contains supplementary material, which is available to authorized users.


Background
Non-communicable diseases (NCDs) cause about 40 million deaths each year globally [1]. The four most important modifiable behavioral risk factors for NCDs include unhealthy diet, physical inactivity, tobacco use [2,3] and harmful use of alcohol [4]. According to the estimates from the most recent Global Burden of Disease study, out of all the deaths due to NCDs in 2015, approximately 12 million deaths were due to unhealthy diet, 6.5 million were due to tobacco use, 1.8 million were due to alcohol and drug use and 1.6 million deaths were attributed to low physical activity [1]. The major risk factors for NCDs are associated with behavioral patterns that are largely established during childhood and adolescence and continue into adulthood [5][6][7]. The onset of many NCDs like diabetes, obesity, and cardiovascular diseases can be prevented or delayed by addressing these risk factors earlier in life [8].
Children and adolescents should be prioritized as target groups for behavioral interventions due to their high adaptability and likelihood to be motivated for appropriate healthy modifications [9]. In support of this, evidence shows that behavioral modifications are more successful if implemented at an early stage [10,11]. Behavioral changes during early years require conducive policies and programs [12]. Hence, in addition to prioritizing children for the adoption of healthy behavioral practices, they should be provided with a supportive environment for behavior change in settings where children live, play and study [13].
Schools are uniquely positioned as ideal settings to model, promote and reinforce healthy behaviors among children and adolescents. Children and adolescents spend much of the daytime at school and can easily access the schools' health-related educational programs. Therefore, schools function as health hubs by educating and imparting healthy habits among students [14,15] as they service a large population of children and adolescents [7]. Evidence suggests that school policies can positively impact Body Mass Index (BMI) [16], physical activity and dietary behaviors [17] among children.
Previous studies have mostly looked at the relationship between school policies and specific risk factors. There exists no review that has systematically identified and collated evidence on the effectiveness of school-based policy interventions collectively for the four major preventable NCD risk factors (unhealthy diet, physical inactivity, tobacco use and alcohol use). Furthermore, no systematic review has examined the impact of school policies on the anthropometric & physiological measures in children. Therefore, the aim of this systematic review was to identify, collate and synthesize the existing literature on the impact of school policies on major risk factors of NCDs.

Search strategy and study selection criteria
A review protocol was developed in accordance with PRISMA guidelines [18]. The search strategy aimed to identify published articles on the effectiveness of school level policy interventions to reduce major preventable risk factors for NCDs (unhealthy diet, physical inactivity, tobacco use, alcohol use, excess body weight, high blood pressure, adverse lipid profile as well as anthropometric and physiological measures) among students. The search was carried out in three electronic databases: Ovid Medline, EMBASE, and Web of Science. The search strategy used to identify the studies in Medline is included in Additional file 1.
The databases were searched for studies published from January 1990 to January 2014. The inclusion criteria were established to include studies assessing effectiveness of either existing or new school based policy interventions among children between the age of 6 to 18 years aimed at the reduction of NCD risk factors. Studies that assessed the effectiveness of pre-school policy intervention were excluded. The detailed inclusion and exclusion criteria guiding the selection of studies for the review is described in Table 1. Duplicate references were removed using software (Endnote X7), and titles and abstracts were independently screened by two reviewers (AS and SB). Any disagreements were resolved by discussion and consultation with a third investigator (MA). Following this step, full text of the selected studies were retrieved and then reviewed for relevance to the inclusion and the exclusion criteria by AS and SB independently. Disagreements at this stage were resolved through discussion between the two reviewers.

Data extraction
Two reviewers (AS and SB) independently extracted information from the selected papers using pre-piloted data extraction forms. Any disagreements were resolved either by discussion or by the intervention of another investigator (MA). The following data were extracted: study characteristics (primary author, year of publication, study setting, age group, sub-groups analysed, sample size, data collection methods, inclusion criteria, randomization information, statistical analysis); intervention or policy component, study outcomes (primary outcomes: BMI, waist circumference, overweight, obesity, physical activity, tobacco use, alcohol use, other relevant outcomes; secondary outcomes: knowledge and attitude), type of effect estimates, main result and statistical significance of differences.

Quality assessment
All the papers included in the review were independently assessed for methodological quality using the Effective Public Health Practice Project (EPHPP) quality assessment tool [19] by AS and then cross-checked by SB. The EPHPP tool contains eight different components but the scoring on quality assessment is done by six parameters. These include selection bias, study design, identification and treatment of confounders, blinding, data collection methods and withdrawals and dropouts. The components were rated strong, moderate, or weak according to a standardized guide and corresponding guidelines in the dictionary. Those with no weak ratings and at least four strong ratings were considered 'Strong.' Those with less than four strong ratings and one weak rating were considered 'Moderate.' Finally, those with two or more weak ratings were considered 'Weak.' The two remaining components within the quality assessment included in the assessment were the integrity of the intervention and the use of appropriate analysis [19].

Synthesis of evidence
Due to the heterogeneity in policy components of the interventions included, outcomes and effect measures, a metaanalysis was not considered appropriate. A description of effectiveness measures and a narrative review were considered appropriate to present the findings of the study.

Results
Overall, 27 studies were included in the review after the full-text screening of the identified articles through systematic database searching (n = 2633), title and abstract screening (n = 90), application of inclusion and exclusion criteria (n = 39) and full-text review (Fig. 1).
The majority of included studies were from highincome countries, USA (15), Australia (4), UK (2), Canada (2), Spain (1), Greece (1), combined USA and Australia (1) with the exception of only one study from India (1). There were 15 interventional studies (eight randomized controlled trials (RCTs), seven quasiexperimental studies), 11 observational studies (ten cross-sectional studies, one case-control) and one natural experiment. Five out of 27 studies were based in schools from socioeconomically deprived areas. Of the 27 studies, ten assessed the effectiveness of multiple policy interventions and 14 studies evaluated multiple outcomes (physical measures, biomarker levels and Table 1 Inclusion and exclusion criteria   Inclusion Criteria  Exclusion Criteria Population Children or adolescents between the age-group of 6 -18 years.

Population
Children or adolescents not in the specified age-group and studies conducted on animal models.

Intervention
Policies that modify the four identified risk factors (unhealthy diet, physical inactivity, alcohol and tobacco use) and associated health related behaviors amongst the students either alone or as part of any intervention program.

Intervention
Policy components those are insufficiently described to enable replication. School policies focusing on differently abled students.

Context
Schools as a setting.

Context
Community, pre-schools and clinical settings.

Outcome
Prevalence of health related behaviors identified as risk factors for NCDs.
Study design Editorials, library thesis, opinions and letters, papers with insufficient methodological details reported to allow critical appraisal of study quality, studies not in English language. Studies published before 1990. Study Design Any experimental or observational study design (randomized controlled trial controlled before-after study, quasi-experimental, interrupted time series, cohort study or cross-sectional study). Fig. 1 Flowchart for study identification and selection process behaviors). The remaining three studies only assessed single intervention or outcome. Collectively, the children within the selected studies ranged from 6 to 17 years and were in grades from 1st to 12th. Apart from one study [20] which included only boys, remaining studies included both boys and girls. Based on the quality assessment of the selected studies, 18 were categorized as having weak methodological quality, six with moderate quality and three with strong methodological quality (Table 2; Additional file 2).

Physical and anthropometric measures
Seven studies assessed the effectiveness of policy interventions or its association with changes in anthropometric measures [14,[21][22][23][24][25][26]. Three studies assessed the effectiveness of school policy in controlling blood pressure [22,26,27]. The policy interventions targeted at anthropometric measurements (BMI, waist circumference, height and weight status) included school nutrition policy initiative [14,21], comprehensive legislation at state level to combat obesity [25], brisk walking lessons [22], teacher trainings, developing activities related to food habits and/ or physical activity [23], fitness guidance, fitness and school nutrition, school-based nutrition, school and home nutrition and home-based nutrition [26]; multicomponent workbooks covering dietary issues, physical activity and fitness [24] and integration of health promotion in the existing curriculum [27] (Table 3).
Mixed results were reported for the effects of school policies on BMI. Non-significant differences or associations with BMI for policy interventions were reported by three studies [14,25,26]; while two studies reported significantly lower progression of BMI among those exposed to policy interventions compared to those who did not [23,24]. The studies that showed policies to be effective in reduction of BMI were assessed to have moderate and strong methodological quality. These effective policy interventions included teacher training, developing activities related to food habits and/or physical activity, multicomponent workbooks covering dietary issues, physical activity, and fitness. Decreased levels of elevated waist circumference as a result of brisk walking lessons was reported in one of the studies [22]; however, this study scored weak in quality assessment. Studies where case definitions included overweight and obesity also showed mixed results. While policy intervention of school nutrition policy initiative was effective in reduction of overweight in one study [14], an increase in the prevalence of both overweight and obesity was observed in another [21]. Though, the study showing effectiveness of school nutrition policy initiative had strong methodological quality.
Two studies that assessed the effectiveness of policies including brisk walking lessons and fitness guidance, fitness and school nutrition, school-based nutrition, school and home nutrition and home-based nutrition in BP control showed desirable effects [22,26]. On the other hand, one assessing the effectiveness of integration of health promotion in the existing curriculum reported non-significant changes [27] (Table 3). However, the three studies were judged to be of weak methodological quality.

Biomarkers
Two out of the 27 studies assessed changes in biomarker levels [22,24]. One study [22] assessed whether extended brisk walking lessons as a school level intervention resulted in changes in serum levels of triglycerides, high-density lipoprotein cholesterol, high-density lipoprotein to total cholesterol ratio and glucose. They reported significantly lower levels of triglycerides, improvements in high-density lipoprotein cholesterol, high-density lipoprotein to total cholesterol ratio and reduction in glucose levels to be associated with the intervention. Similarly, desirable serum level lipid changes were reported by Manios et al. [26], in their study on the effectiveness of multicomponent workbooks covering dietary issues, physical activity and fitness, dental health hygiene, smoking and accident prevention as school level policies. Additionally, teaching aids including posters, audio-taped fairy tales for classroom use, workbooks, and teaching manuals were provided to class teachers and physical education (PE) instructors ( Table 3). The studies were judged to have weak to moderate methodological quality.

Unhealthy diet
The majority of selected studies (n = 15) assessed the effectiveness of policy interventions in changing unhealthy dietary behaviors. These policies ranged from removal of sugar-sweetened beverages (SSBs) and junk food [28][29][30]; change in canteen policies (increasing the availability of lower-fat foods in cafeteria's à la carte areas and implementing school-wide, student-based promotions of these lower-fat foods) [31]; school selfassessment; nutrition education; nutrition policy (meet nutritional standards based on Dietary Guidelines for Americans); social marketing; and parent outreach [14]; fruit truck shops [32]; nutrition education and gardening program [33]; brisk walking lessons [22]; integration of health promotion in curriculum [27]; teacher trainings and development of activities related to food habits and/ or physical activity [23]; fitness guidance, fitness and school nutrition, school-based nutrition, school and home nutrition and home-based nutrition [26]; modified school lunches, enhanced nutrition education and increased opportunities for physical activities [34]; comprehensive school health education, physical education and physical activity, school nutrition and food services, health promotion and wellness, school    counselling, physical and behavioral health services, school climate, physical environment, youth, parent, family and community involvement [35]. Six out of seven studies assessing policy effectiveness in reduction of sugar intake reported desirable changes and reduction in sugar or SSBs consumption [21,26,28,29,35,36]. These effective policies included elimination of SSB and other junk food in schools' food policy, having a school food and nutrition policy in place, school district SSB policies, school nutrition and food services, nutrition-based standards and fitness guidance, fitness and school nutrition, schoolbased nutrition, school and home nutrition and homebased nutrition. Among these policy interventions, studies with moderate methodological quality evaluated school food and nutrition policy and school district SSB policies, while the remaining studies were judged to have weak methodological quality. School policies were also observed to be effective in increasing fruit and vegetable intakes in four out of five studies [23,27,[31][32][33]. Desirable effects of increased fruit and vegetable intakes were noted with the policy interventions of integration of health promotion in the curriculum, change in canteen policies, nutrition education and gardening program, teacher training and development of activities related to food habits and/or physical activity and fruit truck shops. Out of these effective interventions, teacher training and development of activities related to food habits and/or physical activity and fruit truck shops were observed to be reported from studies with moderate and strong methodological quality. Regarding fat reduction and salty snacks, school dietary policies were reported to reduce their prevalence [26,30,36] (Table 4). All the three studies were judged to have weak methodological quality.

Tobacco and alcohol use
Seven out of 27 studies assessed the effects of school tobacco control policies on the prevalence of tobacco use [20,35,[37][38][39][40][41]. These school level tobacco control policies included comprehensive smoking bans, policy orientation towards abstinence and harm minimization principles, penalty on possession of tobacco products among students [37]; school-based smoking intervention: curriculum, parent, nurse counselling cessation support and policy components such as scale for prohibition, strength and characteristics of enforcement [38]. One study assessed seven policy components: developing, overseeing and communicating the policy, purpose, and goals, prohibition, the strength of enforcement, tobacco use prevention education and assistance to overcome tobacco addictions (perceptions regarding policy) [39]. Others assessed a policy banning smoking in school property [40], tobacco-free school policy [41], reduced tobacco promotion and availability around schools [20] and finally comprehensive school health education, school counselling, physical and behavioral health services, school climate, physical environment, youth, parent, family and community involvement [35] (Table 5).
Current smoking was the preferred outcome of evaluation for tobacco use among four out of seven studies [20,37,38,40] and smokeless tobacco use was measured as an outcome in only one of the seven studies [20]. Several other outcomes such as frequency of tobacco use, perception about school smoking and occasional smoking were also assessed in some of the included studies. Two studies reported non-significant differences between those exposed and not exposed to policy [35,37] while five studies reported a significant reduction in tobacco use among those exposed [20,[38][39][40][41]. Among the effective interventions, only school based harm minimization smoking intervention was observed to be tested within a study with strong methodological quality. The remaining four studies scored weak in quality assessment. Only one study tested the association between school-level policies and alcohol use and reported that when the students believed the policy enforcement was not strict, the chances of students consuming alcohol on school grounds were higher [42] (Table 5). This study was judged to have moderate methodological quality.

Physical inactivity
Among the 27 studies, ten assessed associations between school policies and changes in physical activity [22-24, 26, 27, 34, 35, 43-45]. The school policies included learning landscape program (renovation of school grounds) [43]; 90 min moderate intensity physical activity delivered as part of academic instruction [44]; lessons on brisk walking [22]; district mandated physical activity policy (20 min per day) [45]; integration of health promotion in existing curriculum (provision of cognitive behavioral components of health knowledge, health promotion concepts, nutrition and exercise) [27]; teacher trainings, developing activities related to food habits and/or physical activity [23]; fitness trainings [26]; increased opportunities for physical activities (installing physical fitness stations in each classroom; initiating a non-competitive incentive system based on students' personal goals; training of PE teachers and lesson plans for PE teachers) [34]; comprehensive school health education, including physical education and physical activity, school nutrition and food services, health promotion and wellness, school counselling, physical and behavioral health services, school climate, physical environment, youth, parent, family and community involvement [35]; multicomponent workbooks covering dietary issues, physical activity and fitness, dental health hygiene, smoking and accident prevention [24]. All studies reported significant and positive changes in physical     activity with the implementation of school policies except one [34] (Table 6). However, majority of these studies were of weak methodological quality and policies that were observed to be effective from moderate and strong methodological studies included teacher trainings, developing activities related to food habits and/or physical activity; and, multicomponent workbooks covering dietary issues, physical activity and fitness, dental health hygiene, smoking and accident prevention.

Policy effectiveness
In total, for 103 outcomes independently evaluated within these studies, 47 outcomes (46%) had significant desirable changes when exposed to the school policies. In terms of the frequency, these interventions included: school selfassessment (n = 1); nutrition education (n = 5); nutrition policy (n = 9); social marketing and parent outreach (n = 1); health education (n = 5); extended brisk walking lessons (n = 6); teacher trainings (n = 3); guidance around fitness or fitness training (n = 4); school district SSB policies (n = 1); integration of health promotion in curriculum (n = 2); creating opportunities for physical activity (n = 1); smoking bans, policy orientation towards abstinence and penalty on possession of tobacco (n = 1); school-based harm minimization intervention (n = 2); district or school-based tobacco control policies (n = 3); tobacco free school policy (n = 1); alcohol abstinence and harm minimization messages (n = 2); investment on school infrastructure to promote physical activity (n = 2); physical activity in curriculum (n = 1); district mandated physical activity policy (n = 1). However, the policy interventions observed to be effective after restricting the evidence from studies having strong and moderate methodological quality were fewer in numbers. These included teacher trainings, activities related to food habits and physical activity, and multicomponent workbooks for desirable outcomes on anthropometric measurements, and biomarkers. For the outcome of change in dietary behaviors, evidence was supportive for school district SSB policies, school food and nutrition policy. Evidence from strong and moderate methodological quality studies showed that school based harm minimization smoking intervention was effective for reduction of smoking, and low policy enforcement for alcohol use for the desirable outcome of reduced alcohol intake. Finally, for desirable changes in physical activity; the teacher trainings, developing activities related to food habits and/or physical activity; and, multicomponent workbooks were reported to be effective.

Discussion
The current study systematically reviewed the evidence on the effectiveness of school policies in the reduction of risk factors for NCDs. Although the overall evidence indicates effectiveness of policies on behavioural outcomes and biomarkers, majority of these studies were judged to have weak to moderate methodological quality. Compared to these outcomes, school policies were evident to be less effective for the outcomes such as anthropometric measurements, overweight and obesity, and, alcohol use. None of the included studies reported long-term follow-up of participants. Therefore, it is not known whether beneficial changes in NCD risk factors resulting from school policy interventions are sustained in the longer term. The NCD risk factors evaluated within the selected studies ranged from individual health behaviors to anthropometric measurements and biomarkers. The observed differences in the effectiveness of the policies according to these different types of outcomes may reflect the lag time between exposure to the intervention and effect on the outcome (NCD risk factor). For example, behavioral changes such as reduced sugar intake may be immediate following an intervention, but behaviors need to be sustained over a longer period to produce changes in anthropometric measures. These differences can also be due to the variations in the number of studies reporting these outcomes. While most selected studies tested the effectiveness of school policies on health behaviors (diet, smoking and physical activity) and anthropometric measurements, fewer studies tested associations between the school policies and biomarkers as well as between school policies and alcohol use.
High sugar consumption is associated with multiple NCDs including overweight/ obesity [46], diabetes [47] and dental caries [48]. The effective policy interventions targeted at the school level to reduce sugar consumption among school children were to have a school nutrition policy restricting SSBs. These findings are consistent with those reported in independent reviews on determinants of high sugar intake [49,50]. Tobacco use is a well-established independent risk factor for NCDs and the most prevalent behavioral risk factor globally [51]. The most efficient school policy interventions included school-based harm minimization smoking intervention. These findings further substantiate the results on the positive influence of school environment on tobacco abstinence among children [52,53]. Physical inactivity or sedentary behavior is also associated with worse health outcomes [54][55][56]. Teacher trainings, developing activities related to food habits and/or physical activity and, multicomponent workbooks were reported to be effective in increasing physical activity among school children from methodologically stronger studies. Another systematic review on school policy interventions and NCD risk factors reported the majority of the school policy interventions being effective [57]. This systematic review had several strengths and some limitations. A previous systematic review on the effectiveness of school policies was limited to the outcomes like physical inactivity, diet, and tobacco consumption [57] while the current review also included anthropometric measures, alcohol use, and biomarkers. Furthermore, the current systematic review included different study designs. Evidence suggests that RCTs are inappropriate for evaluating most health promotion interventions [58]. Thus, being inclusive of studies regardless of study designs allowed to present more comprehensive evidence on school policies. The review also had some limitations. First, the review was limited to Englishlanguage articles, which may have led to the exclusion of some relevant studies in other languages. Second, the review did not include grey literature including government reports due to lack of access. However, governments are likely to evaluate context specific school policy interventions which are less likely to be generalized to different contexts. Third, a meta-analysis of the evidence could not be conducted due to the heterogeneity among the policy components of the interventions included and the lack of uniformity in outcomes and effect measures. Finally, the evidence can only be summarized for the direction of association rather than its strength due to multiple heterogeneities within studies in terms of exposure, outcomes, and the effect estimates.

Research and policy implications
The findings from the current review have important research and policy implications. The mixed evidence on the effectiveness of school policies according to different outcomes indicates that when planning intervention studies, the time lag between any policy exposure and the outcomes of interest should be considered carefully. Another important consideration is the duration of follow-up within studies; studies with longer-term follow-up of students are needed to assess whether positive changes in NCD risks are maintained beyond the intervention period and school years. Some of the school based policy interventions are expensive to design and resource intensive to carry out. The lack of good quality evidence on the effectiveness of school based policy interventions highlights the need for well-designed studies to inform the policymakers. Finally, all except one study were from high-income countries, which underscores the gap in research evidence from low-and middleincome countries. Global estimates indicate that 80% of the deaths due to NCDs occurs within low-and middleincome countries [59]. Thus more research on the effectiveness of school policies in the reduction of NCDs and their risk factors should also be conducted within these countries.
In most societies, NCDs are unequally distributed [60] and a shift in policy attention from individual to environmental and structural factors, such as the school environment, could be a more equitable approach [61]. The evidence from this systematic review will be used as a wider framework to aid in developing a contextually relevant and tailored comprehensive NCD intervention model for schools.
It will also guide the drafting of an evidence-based global school policies checklist for promoting a healthy lifestyle and preventing NCDs. The school checklist will inform policy-makers on aligning school curriculum, school activities and school health services, food availability and school infrastructure to be conducive for NCD prevention.
the study protocol, assisted with data searches, provided critical intellectual inputs for revising and finalizing the manuscript. MP is a co-investigator for this study from the UK. She contributed towards concept and design of the study, reviewed and provided technical guidance on study related documentation including all aspects of this manuscript. She also provided final approval to the manuscript. SK is a co-investigator for this study and the senior technical guide from the UK. He contributed substantially towards the study concept and design. He along with MA provided final approval to study protocol and related documents, including this manuscript and revised it critically for intellectual content. MA is the principal investigator for this study. She contributed substantially towards the concept and design of the study. She was also the reviewer 3 who resolved any discrepancies between findings of reviewers 1 and 2. She provided overall guidance and final approval to all study related documents including this manuscript, provided critical inputs for revising the manuscript and is accountable towards all aspects of this study along with other investigators. All authors read and approved the final manuscript.