Physical Activity–Related Policy and Environmental Strategies to Prevent Obesity in Rural Communities: A Systematic Review of the Literature, 2002–2013

Introduction Health disparities exist between rural and urban residents; in particular, rural residents have higher rates of chronic diseases and obesity. Evidence supports the effectiveness of policy and environmental strategies to prevent obesity and promote health equity. In 2009, the Centers for Disease Control and Prevention recommended 24 policy and environmental strategies for use by local communities: the Common Community Measures for Obesity Prevention (COCOMO); 12 strategies focus on physical activity. This review was conducted to synthesize evidence on the implementation, relevance, and effectiveness of physical activity–related policy and environmental strategies for obesity prevention in rural communities. Methods A literature search was conducted in PubMed, PsycINFO, Web of Science, CINHAL, and PAIS databases for articles published from 2002 through May 2013 that reported findings from physical activity–related policy or environmental interventions conducted in the United States or Canada. Each article was extracted independently by 2 researchers. Results Of 2,002 articles, 30 articles representing 26 distinct studies met inclusion criteria. Schools were the most common setting (n = 18 studies). COCOMO strategies were applied in rural communities in 22 studies; the 2 most common COCOMO strategies were “enhance infrastructure supporting walking” (n = 11) and “increase opportunities for extracurricular physical activity” (n = 9). Most studies (n = 21) applied at least one of 8 non-COCOMO strategies; the most common was increasing physical activity opportunities at school outside of physical education (n = 8). Only 14 studies measured or reported physical activity outcomes (10 studies solely used self-report); 10 reported positive changes. Conclusion Seven of the 12 COCOMO physical activity–related strategies were successfully implemented in 2 or more studies, suggesting that these 7 strategies are relevant in rural communities and the other 5 might be less applicable in rural communities. Further research using robust study designs and measurement is needed to better ascertain implementation success and effectiveness of COCOMO and non-COCOMO strategies in rural communities.


Introduction
Health disparities exist between rural and urban residents; in particular, rural residents have higher rates of chronic diseases and obesity. Evidence supports the effectiveness of policy and environmental strategies to prevent obesity and promote health equity. In 2009, the Centers for Disease Control and Prevention recommended 24 policy and environmental strategies for use by local communities: the Common Community Measures for Obesity Prevention (COCOMO); 12 strategies focus on physical activity. This review was conducted to synthesize evidence on the implementation, relevance, and effectiveness of physical activity-related policy and environmental strategies for obesity prevention in rural communities.

Methods
A literature search was conducted in PubMed, PsycINFO, Web of Science, CINHAL, and PAIS databases for articles published from 2002 through May 2013 that reported findings from physical activity-related policy or environmental interventions conducted in the United States or Canada. Each article was extracted independently by 2 researchers.

Results
Of 2,002 articles, 30 articles representing 26 distinct studies met inclusion criteria. Schools were the most common setting (n = 18 studies). COCOMO strategies were applied in rural communities in 22 studies; the 2 most common COCOMO strategies were "enhance infrastructure supporting walking" (n = 11) and "increase opportunities for extracurricular physical activity" (n = 9). Most studies (n = 21) applied at least one of 8 non-COCOMO strategies; the most common was increasing physical activity opportunities at school outside of physical education (n = 8). Only 14 studies measured or reported physical activity outcomes (10 studies solely used self-report); 10 reported positive changes.

Introduction
Rural residents have higher rates of chronic diseases and obesity than urban residents (1-5). Rural residents may have as much as 6.2% higher prevalence of obesity than urban residents (6,7). Physical inactivity is associated with higher rates of chronic diseases and obesity (7,8), and some research suggests that rural residents are less physically active than urban residents (9-11). Rural residents also have higher rates of poverty, fewer community resources, less access to preventive services and health care, greater geographic dispersion, and more transportation challenges (eg, lack of public transit, greater travel distance) than urban residents (12)(13)(14)(15)(16)(17)(18). Sixteen percent of Americans live in rural areas that encompass 72% of land in the United States. Although evidence supports the effectiveness of policy and environmental strategies to prevent obesity and promote health equity, much of this evidence is derived from nonrural settings (13,19,20).
In 2009, the Centers for Disease Control and Prevention (CDC) recommended 24 strategies for local communities to use in planning and monitoring obesity-related policy and environmental changes using preexisting data sources: the Common Community Measures for Obesity Prevention (COCOMO) (21). Twelve strategies focus on physical activity (PA) ( Table 1): 4 strategies to "encourage physical activity or limit sedentary activity among children and youth" (strategy nos. 12-15) and 8 strategies to "create safe communities that support physical activity" (strategy nos. [16][17][18][19][20][21][22][23]. The purpose of this study was to conduct a systematic literature review of the implementation, relevance, and effectiveness of physical activity-related policy and environmental strategies for obesity prevention in rural communities, including both COCOMO and non-COCOMO approaches. A secondary aim was to synthesize the evidence on the implementation success of the 12 physical activity-related COCOMO strategies in rural communities.

Data sources
A literature search was conducted in the following databases: PubMed, PsycINFO, Web of Science, Cumulative Index to Nursing and Allied Health Literature (CINHAL), and Public Affairs Information Service (PAIS). The search included articles published in English from 2002 through May 2013 and focused on findings from PA-related policy or environmental interventions. Each search used the following terms: "rural" AND "physical activity or exercise or sedentary or inactivity" AND "community or environment or policy." Searches were repeated in a secondary literature search using search terms for Native American communities ("tribal" OR "reservation" OR "Native American" OR "indigenous") and predominantly rural states. "Predominantly rural states" were identified where 1) most (half or greater) of the state was identified as rural using the Rural-to-Urban Continuum Codes and the Office of Management and Budget maps or 2) substantial portions of a state were identified as frontier using the Rural Assistance Center's Frontier map (22,23). The following states were designated as predominantly rural: Alaska, Idaho, Kansas, Maine, Mississippi, Montana, Nebraska, Nevada, New Hampshire, New Mexico, North Dakota, Oklahoma, Oregon, South Dakota, Texas, Utah, Vermont, West Virginia, and Wyoming. Relevant references cited in each identified study were also included in the secondary literature search. Methods mirrored a companion literature review describing application of nutrition-related COCOMO strategies in rural communities (24).

Study selection
At least 2 researchers reviewed titles, abstracts, and complete texts of articles for inclusion. Studies were included that reported findings from empirical formative, process, or outcome research with strategies aimed at changing policy or environments to support PA in rural US or Canada communities. Publications were excluded if 1) both rural and urban communities were included and rural-specific findings were not reported, 2) the primary focus was on instrument development or individual-level behavioral change, or 3) studies were descriptive or did not describe an intervention.

Data extraction
The article extraction team consisted of 18 trained researchers. Data for each article were extracted independently by 2 trained researchers. We used a customized Qualtrics (Qualtrics LLC) online survey as a tool to extract information about study setting, geographic location, sample characteristics, rural definition, design, methods, results, and bias-risk assessment (25)(26)(27). After independent extraction, results were compared and discrepancies were resolved by consensus. Study quality was examined for randomized control trials (RCTs) using Cochrane Collaboration's assessment tool. We used GRADE guidelines of bias risk for observational studies to assess non-RCTs, including formative studies, because the Cochrane tool focuses only on . The Cochrane tool assesses risk of bias across 6 categories: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias (25,26); all categories were assessed as designed. GRADE guidelines assess risk of bias across 4 categories: appropriate eligibility criteria, measurement of exposure and outcome, control of confounding, and incomplete follow-up (27); all categories were assessed as designed. Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs: low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). Extraction data entered into Qualtrics were downloaded into Excel for synthesis. We organized data into the following categories: 1) study location, setting, approach, and bias-risk assessment; 2) COCOMO strategies used; 3) non-COCOMO strategies used; 4) measurement of policy and environmental strategies; and 5) intervention effects on policy, environment, behavioral, and health outcomes.

Results
Searches returned 9,879 articles, of which 2,002 were identified as relevant for further screening based on title and abstract. Duplicates were removed, leaving 488 records for full-text screening; 443 of these did not meet inclusion criteria. The remaining 45 articles represented 41 distinct studies and were assigned for data extraction; 15 articles were excluded during extraction for various reasons (Figure). Thus, 30 articles representing 26 distinct studies were included in the final synthesis.

Study location, setting, approach, and bias-risk assessment
Of the 26 studies, 3 were conducted in Canada and 23 in the United States; 4 studies were conducted with American Indian tribes or First Nations of Canada (Table 2). Rural location of 19 studies was defined by authors as "rural," "small town," or "remote"; 4 studies provided information about population density to define rurality, and 3 were identified as rural only through descriptions of tribal or reservation areas. Study settings included schools (n = 18), communities (defined broadly without identification of an entity, organization, or institution; n = 12), worksites (n = 5), churches (n = 1), homes (n = 2), and childcare (n = 1); 5 interventions targeted multiple settings. In the 18 school-setting studies, interventions resulted in changes that affected students (n = 14), changes in the use of facilities for the community (n = 3), or changes that affected employees (n = 1). Study designs included formative (n = 7), process (n = 16), or outcome (n = 20) evaluations; 15 included 2 or more types of evaluation. Only 3 studies were RCTs. Bias-risk assessments showed that 19 studies had high risk of bias, 4 studies had medium risk, and 3 studies had low risk ( Table 3). None of the 23 non-RCTs adequately controlled for confounding, and 5 non-RCTs had flawed measurement. Six non-RCTs developed and applied appropriate eligibility criteria, and 7 non-RCTs had complete follow-up. Of the 3 RCTs, one had medium risk and 2 had high risk of bias (Table 3). Sequence generation was absent in all RCTs, and all reported selective outcome data and had other sources of bias.

Use of COCOMO strategies
Although only one study referenced CDC's COCOMO strategies (48), 22 of the 26 studies applied at least one PA-related COCOMO strategy (Table 2), and 4 studies did not apply any COCOMO strategies. The mean number of COCOMO strategies applied was 2.0 (standard deviation [SD], 2.3; range, 0-12). The 2 most commonly applied COCOMO strategies were no. 18, "enhance infrastructure supporting walking" (n = 11), and no. 14, "increase opportunities for extracurricular physical activity" (n = 9). Fourteen studies applied at least one COCOMO strategy to "encourage physical activity or limit sedentary activity among children and youth," 12 studies applied at least one COCOMO strategy to "create safe communities that support physical activity," and 4 studies applied at least one of each. Identified barriers to implementing these strategies in rural areas were staff turnover and lack of staff buy-in, organizational or community support, resources, and political will. Identified facilitators were communication, accountability, training, and ease of implementation.

Use of non-COCOMO strategies
One or more non-COCOMO strategies were mentioned in 21 studies (Table 2). Four studies incorporated only non-COCOMO strategies and 6 studies incorporated 2 or more non-COCOMO strategies. The mean number of non-COCOMO strategies applied was 1.1 (SD, 0.8). Eight non-COCOMO strategies were identified: increase PA opportunities at school outside of physical education (PE) (n = 8) (eg, classroom activity breaks, longer school recess, reversing lunch and recess); increase amount of and access to PA equipment or improve existing equipment resources (n = 6); promote PA resources (n = 6) (eg, signs to promote walking routes or trails); provide access to public buildings after hours for PA purposes (n = 3); adopt worksite policies or practices (n = 3) (eg, allowing PA breaks during workday); reduce home screen time (n = 1); reduce school or preschool sedentary time (n = 1); and school district-wide adoption of a PA-supportive curriculum (n = 1). The mean number of COCOMO and non-COCOMO strategies applied per study was 3.1 (SD, 2.3).

Measurement of policy and environmental strategies
Measurement of policy or environmental changes was not consistent or standard across studies, and researchers often did not use COCOMO-suggested measurements (Table 3). For example, studies (n = 6) that "increased the amount of physical activity in PE programs in schools" (strategy no. 13) documented results by indicating use of a modified PE program, increased minutes in PE or increased time in 43,53,57). Some studies used a similar non-COCOMO metric to measure change. For example, the 8 studies that "increased opportunities of extracurricular physical activity" (strategy no. 14) measured change by indicating the presence of increased opportunities for PA (28,39,42,43,45,47,(53)(54)(55).
Intervention effects on policy, environment, behavioral, and health outcomes Sixteen interventions had at least one positive environmental change or result, and 18 interventions reported a positive policy change or result (Table 3). Three studies focused solely on formative approaches without reporting policy or environmental results ( Table 4). Seven of the 8 nonformative studies that "required PE" (strategy no. 12) or "increased amount of physical activity in PE programs" (strategy no. 13) reported a positive PE policy or environmental change (Table 3). All 3 studies that adopted worksite policies promoting PA documented policy implementation, and 2 studies measured improvements in health status (28,38).

Discussion
We found 26 unique studies that implemented PA-related COCOMO or non-COCOMO strategies in rural communities. Given the variation in settings, methods, and results of the studies reviewed, we were unable to empirically assess effectiveness; however, these findings provide a synthesis of current practices and guidance on implementing policy and environmental strategies in rural communities.
Seven of the 12 PA-related COCOMO strategies (nos. 12-18) were applied in 2 or more nonformative studies, suggesting that these strategies are relevant in rural communities. All but 2 studies (29,47) that used these 7 strategies reported effectively implementing them in the target rural communities. Ten studies reported improvements in PA after implementation of policy or environmental changes. "Enhancing infrastructure supporting walking" (no. 18) was implemented in 6 of these 8 studies, with significant changes in 4 studies. However, because 5 of these 6 studies implemented more than one strategy, we cannot attribute the improvement in PA to this strategy alone. Three of the COCOMO strategies were not implemented in any of the reviewed studies, and 2 strategies were implemented in only one study each, suggesting these strategies might be less relevant for rural communities, as originally cautioned when the guidelines were released (21). These strategies relate to location of schools, improvement of public transportation, mixed-use zoning, enhanced personal safety, and traffic safety in areas where people could be physically active. Rural communities may not use these strategies because they lack the resources to implement these strategies or because of other constraints related to small and dispersed populations in comparison with urban and suburban communities. For example, many rural areas have limited or no public transportation systems (59,60) and may not have the tax base or concentration of users to make a public transportation system feasible (61). The studies reviewed implemented 8 non-COCOMO strategies. Although these strategies may not be germane to rural areas only, they have been implemented in rural areas; more research on their effectiveness in rural areas is warranted. Given the increase in policy and environmental approaches for improving PA after the publication of the COCOMO strategies in 2009, our review is beneficial to the field and indicates it may be an opportune time to update PA-related COCOMO strategies (21).
Most policy or environmental strategies implemented in the studies reviewed focused on schools, whether the target population was students, school employees, or community members using the school facilities outside of school time. Use of schools as the focal point for obesity-related interventions aligns with Institute of Medicine recommendations (68). In rural areas where community resources and safe places to be active are limited (59,60,69-71), school facilities and resources (eg, gyms, fields, playgrounds) are often some of the few, if not the only community assets for PA (72). However, many rural areas are consolidating their school districts and building new schools on the outskirts of rural communities or on state highways rather than renovating existing schools or building new schools within municipal domains or current residential areas; this trend may create school grounds that are less ac- www.cdc.gov/pcd/issues/2016/15_0406.htm • Centers for Disease Control and Prevention cessible for the more populous areas of a rural county (60,71,73,74). Thus, when school-consolidation decisions are made, accessibility of school facilities for PA should be considered.

PREVENTING CHRONIC DISEASE
Recommended COCOMO measurements were not used in the studies reviewed, suggesting that COCOMO measurement approaches might need to be adapted for rural areas. For example, for strategy no. 18, "enhance infrastructure supporting walking," the suggested COCOMO measurement is miles of paved sidewalks relative to total street miles. In the studies reviewed, enhancing infrastructure for walking included building walking trails and paving sidewalks; thus, miles of paved sidewalks would not capture all possible supportive infrastructure changes. The scale of some suggested COCOMO measurements are too large to be pertinent to smaller communities. For example, the suggested COCOMO measure for no. 13, "increase the amount of physical activity in PE programs in schools," is whether the largest school district in the local jurisdiction has a policy that requires K through 12 students to be physically active for at least 50% of PE time. Small school districts, common in rural areas, may not be able to provide the level of detail necessary to determine success using this metric, and rural communities may have only one or 2 school districts; an appropriate rural metric could be the percentage of schools in a district that require K through 12 students to be physically active for at least 50% of PE time. Creating standard valid, reliable, specific, appropriate, and feasible metrics for policy and environmental strategy measurement for rural communities would help these communities better assess the success of policy and environmental strategies and help build an evidence base in rural communities.
Measurement of PA outcomes in the studies reviewed was rare and lacked consistency and methodological strength, limiting interpretation. When PA change was reported, most studies used a form of self-report. Few studies used objective measurement, and those that did measure PA objectively only did so in a subset of their sample, with half using pedometers. Accelerometers are a valid, reliable, and practical measure of PA and are used nationally and internationally (75). Rural evaluations need to consistently measure PA across studies using accelerometers to allow for better understanding of intervention effectiveness and comparison across the urban-rural continuum. Because of the decreasing costs of accelerometers and the ability to borrow units or purchase used units, recent rural community-based approaches have used accelerometers to measure PA and suggested strategies to improve feasibility, accuracy, and consistency (eg, text/email "wear" reminders, data collection methods, scoring methods) for using them in rural communities (75-77).
Lack of detail on study methods and variation in study design, measurement of outcomes, and context limited our ability to compare results of strategies across studies and examine effectiveness. Most studies were biased across assessment categories, indicating overall weakness in research design. Only 14 studies measured change in PA in response to policy or environmental strategies, and measurement approaches greatly varied. Future studies should incorporate elements of strong study design, such as clearly defined inclusion or exclusion criteria, protocols for data collection and use of measurement tools, measurement of potential confounders, reporting of sample size and estimated reach, and objective measurement of change in PA behavior.
Despite the challenges discussed and the challenges inherent in the subjective methods of systematic reviews, this review and its companion (24), provide a synthesis of the data on the use of COCOMO strategies in rural communities. The main findings of both reviews include the importance of making schools the focal point of nutrition-and PA-related interventions and building on existing community resources. Additionally, several nutrition-and PA-related COCOMO strategies, such as improvement of public transportation or geographic availability of supermarkets, may not be applicable to rural communities. We recommend inclusion of non-COCOMO PA-related strategies and refinement of current COCOMO recommended measurements. Improvements for current COCOMO nutrition-related strategies have been suggested (24). Both reviews recommend the use of stronger study design and measurement of policy, environment, and behavior in future studies (24). We echo a conclusion that additional guidance on implementation of policy or environmental strategies in rural communities could be found in research not published in scientific literature (eg, websites, gray literature) (24). Although we used many strategies to identify studies conducted in rural settings, the definition of "rural" varies (71), and studies that were not explicitly identified as rural by their authors might not have been included. Although the variability in rural communities (by geography, population density, topography, resources, and other factors) should be considered in obesity prevention approaches (71), these reviews described strategies that were successfully implemented in multiple rural communities.
COCOMO strategies and recommended measurements provide an evidence-based approach to address obesity and measure the success of intervention strategies. Most PA-related strategies appeared to be applicable in rural communities; however, measurements recommended by COCOMO were not always appropriate. Several non-COCOMO strategies were effectively implemented in rural communities. Generating a database of recommended strategies and measurements relevant to rural communities should be considered. Further research, using robust study designs and PREVENTING CHRONIC DISEASE measures, is needed to better ascertain the success and effectiveness of implementing policy and environmental strategies in rural communities. This information could aid policy makers and community leaders in decision making on resource allocation and obesity-prevention efforts in their rural communities.        -up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.

Tables
(continued on next page)

Psychosocial
Increase in PA self-efficacy (self-report in survey).
Behavior No significant differences in PA change between groups for subset using 1 day of accelerometer data (n = 278), although nonsignificant increases in PA were found for intervention group (accelerometer, TriTrac-R3D); significantly higher self-reported PA at post-test for intervention schools (self-report: 24-hr PA recall survey).
Health status No significant differences between intervention and control groups for all anthropometric variables (objectively measured: BMI; % body fat using bioelectrical impedance; skinfold thickness).
Abbreviations: -, data not reported; BMI, body mass index; BRFSS, Behavioral Risk Factor Surveillance System; HDL, high-density lipoprotein, IPAQ-Short, International Physical Activity Questionnaire-Short Form; LDL, low-density lipoprotein; PA, physical activity; PE, physical education; RCT, randomized control trial; MET, metabolic equivalent, PAQ, Physical Activity Questionnaire for Children and Adolescents; SAPAC, Self-Administered Physical Activity Checklist. a Bias risk was determined using Cochrane Collaboration's assessment tool for RCTs and GRADE guidelines for non-RCTs (25-27). The Cochrane tool assesses risk of bias across 6 categories: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias (25,26); GRADE guidelines assess risk of bias across 4 categories: appropriate eligibility criteria, measurement of exposure and outcome, control of confounding, and incomplete follow-up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.
(continued on next page)  Abbreviations: -, data not reported; BMI, body mass index; BRFSS, Behavioral Risk Factor Surveillance System; HDL, high-density lipoprotein, IPAQ-Short, International Physical Activity Questionnaire-Short Form; LDL, low-density lipoprotein; PA, physical activity; PE, physical education; RCT, randomized control trial; MET, metabolic equivalent, PAQ, Physical Activity Questionnaire for Children and Adolescents; SAPAC, Self-Administered Physical Activity Checklist. a Bias risk was determined using Cochrane Collaboration's assessment tool for RCTs and GRADE guidelines for non-RCTs (25-27). The Cochrane tool assesses risk of bias across 6 categories: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias (25,26); GRADE guidelines assess risk of bias across 4 categories: appropriate eligibility criteria, measurement of exposure and outcome, control of confounding, and incomplete follow-up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.

PREVENTING CHRONIC DISEASE
(continued on next page)

Behavior
Community survey participants showed significant increase in days per week of PA (self-reported in survey, tool not specified); adult PA program participants showed increase in percentage engaged in 30 min per day of PA 3 to 7 days per week after 1 semester (self-reported in survey, tool not specified).
Health status 139 Worksite wellness participants lost on average 3 pounds after 1 semester and 63 participants lost on average 5 pounds after 2 semesters (objectively measured).
Abbreviations: -, data not reported; BMI, body mass index; BRFSS, Behavioral Risk Factor Surveillance System; HDL, high-density lipoprotein, IPAQ-Short, International Physical Activity Questionnaire-Short Form; LDL, low-density lipoprotein; PA, physical activity; PE, physical education; RCT, randomized control trial; MET, metabolic equivalent, PAQ, Physical Activity Questionnaire for Children and Adolescents; SAPAC, Self-Administered Physical Activity Checklist. a Bias risk was determined using Cochrane Collaboration's assessment tool for RCTs and GRADE guidelines for non-RCTs (25-27). The Cochrane tool assesses risk of bias across 6 categories: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias (25,26); GRADE guidelines assess risk of bias across 4 categories: appropriate eligibility criteria, measurement of exposure and outcome, control of confounding, and incomplete follow-up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.
(continued on next page)   -up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.
(continued on next page)   -up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.
(continued on next page)   -up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.
(continued on next page)  Indoor and outdoor walking trails established; instituted "move it moments" (5 min of PA); all teachers reported using "move it moments"; most teachers reported most or all students wore pedometers.

Psychosocial
Teachers perceived "move it moments" improved student behavior during class.

Behavior
Significant increase in steps per day (pedometer).

Health status
No change in BMI z scores in first 7 months of the program; students with healthy weight or at risk for overweight (85th-<95th percentile) were more likely to decrease BMI z score; students who were overweight were more likely to show no change in BMI z score than were healthy-weight or at-risk-foroverweight students.
Schetzina et al, Family & Community Health, 2011 (52) Abbreviations: -, data not reported; BMI, body mass index; BRFSS, Behavioral Risk Factor Surveillance System; HDL, high-density lipoprotein, IPAQ-Short, International Physical Activity Questionnaire-Short Form; LDL, low-density lipoprotein; PA, physical activity; PE, physical education; RCT, randomized control trial; MET, metabolic equivalent, PAQ, Physical Activity Questionnaire for Children and Adolescents; SAPAC, Self-Administered Physical Activity Checklist. a Bias risk was determined using Cochrane Collaboration's assessment tool for RCTs and GRADE guidelines for non-RCTs (25-27). The Cochrane tool assesses risk of bias across 6 categories: sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias (25,26); GRADE guidelines assess risk of bias across 4 categories: appropriate eligibility criteria, measurement of exposure and outcome, control of confounding, and incomplete follow-up (27). Risk of bias was rated as low (score of 1), high (score of 0), or unclear (score of 0) for each Cochrane or GRADE category based on study type (25); overall summary scores for bias risk were calculated and categorized as low, medium, or high (RCTs, low risk = 5 or 6, medium risk = 2-4, and high risk = 0 or 1; non-RCTs: low risk = 3 or 4, medium risk = 2, and high risk = 0 or 1). b When reported, we listed reach, which is the number of community members potentially affected by an intervention. c When reported, we listed the sample size of participants who completed evaluation measures for each study. d When reported, we listed the factors influencing intervention implementation.
(continued on next page)