The ATS interventions make it easier and safer for children to walk and bike to school by targeting the physical or social safety of common routes to school or by promoting safe travel behaviors. Interventions must include ≥1 of the following components, based on the Safe Routes to School model¹⁸:

engineering—improvements to the built environment infrastructure;

This study was conducted using established methods for systematic economic reviews approved by the CPSTF.¹⁹ The team included subject matter experts on physical activity and active travel from various agencies, organizations, and academic institutions, in addition to members of the CPSTF and experts in systematic economic reviews from the Community Guide Office at the Centers for Disease Control and Prevention. Two reviewers independently screened the search yield, abstracted information from the included studies, computed economic estimates, and quality scored each estimate. Disagreements were resolved through discussions.

The present study asks what it costs to implement ATS interventions and what the economic benefits are that result from the intervention. Do the economic benefits due to intervention exceed the cost to implement?

This economic review framework in Figure 1 depicts how the intervention is expected to work and the pathways to economic costs and benefits. Moving from top left to the right, the targeted population includes students and their parents for whom walking or bicycling to school is feasible, plus other community residents who may use the routes for other purposes. All students and parents have multiple mode choices available to them to travel between home and school, including private automobiles, school bus, walking, bicycling, and public transit. The effective intervention leads to an increase in the proportion of students who choose the ATS mode (i.e., walking or bicycling), and a reduction in the proportions using other modes of travel, as was shown in the review of effectiveness.¹⁶ Health improves from the increased physical activity of active travel and averted longer-term diseases associated with inactivity and excess weight. Each travel mode choice has particular private and societal costs that derive from monetization of effects on resource use, travel time, health, traffic-related injuries, and impacts on the environment. Where these costs are reduced because of the intervention are the economic benefits due to intervention. ATS interventions also improve the social environment (e.g., a Walking School Bus program; safety in numbers) and the built environment’s physical safety, thereby reducing injuries for both current and new users of the routes.

The economic costs and conequences of the interventions are shown at the bottom of Figure 1. At the bottom left, economic evaluations of these interventions capture the cost to implement the intervention, which includes planning, infrastructure changes, education, promotion, and enforcement activities. The components marked with asterisks are expected to be drivers of the magnitude of estimates. At the bottom right are the monetized and other benefits due to intervention. The total societal monetized benefit of the intervention is therefore the sum of the following elements of costs associated with all individuals and their travel mode choices post intervention minus the costs at baseline: physical resources and travel time, environmental impacts, near and longer-term healthcare costs, and injuries and fatalities. All the components of benefits are expected to be drivers of the magnitude of the estimate, and are therefore marked with asterisks. The framework in Figure 1 postulates that ATS interventions cause a shift toward cheaper, safer, environmentally friendlier, and healthier ATS modes and away from the use of private automobiles and busing.

Quality assessment of the economic evidence follows methods developed by the Community Guide for systematic economic reviews.¹⁹ In general terms, individual estimates from the studies are assigned a quality score of good, fair, or limited, based on assessments within each of 2 domains. First, quality is assessed based on the domain of capture; that is, how well an economic estimate captures the drivers from among its components. A driver of an estimate is a component that contributes substantially to its magnitude. Second, quality is assessed based on the domain of measurement, which is the appropriateness of methods used by the study to measure and value the estimates. The final quality assignment is the lower of the 2 assigned quality scores. The quality of a composite estimate such as cost benefit is the lower of quality assigned to its individual cost and benefit parts. Limited quality estimates are excluded from the body of evidence.

The quality assessment process just described in general terms was adapted within a quality assessment tool developed for the specifics of the present review, and is available in the Appendix. Within the domain of capture, engineering and education or encouragement were considered drivers of intervention cost. The drivers of benefits were costs of private automobile use, injuries and fatalities, travel time, healthcare cost related to physical inactivity and body weight, and the health and other impacts of congestion, pollution, and greenhouse gases. Note these were the drivers also identified in Figure 1. Within the domain of measurement, the quality of benefit estimates and cost estimates were additionally assessed in the following listed areas along with what are deemed appropriate for the present intervention and review. Limitation points were assigned for departures from what is appropriate.

Opportunity is provided in the assessment process to assign a fatal flaw that automatically scores an estimate as “limited” quality. A fatal flaw is some feature of the estimate that almost certainly causes it to severely misrepresent the true cost or benefit of the ATS intervention.

All monetary values are in 2019 U.S. dollars, adjusted for inflation using the Consumer Price Index,²⁰ and converted from foreign currency denominations using purchasing power parities.²¹ All analyses were conducted during September 2018 through May 2019.

Peer-reviewed and gray literature were searched for economic evaluations. Criteria for inclusion were as follows: met the definition of the intervention, conducted in a high-income country.¹⁷ written in English, and included ≥1 economic outcomes described in the research questions.

A formal search was conducted within PubMed, Scopus, Cochrane, National Transportation Library, National Technical Information Service, and EconLit for papers published through July 2018. Informal searches were also conducted for reports from governments and non-government organizations using the Google and Google Scholar search engines. Finally, citations from another review²² and reference lists in included studies were screened and subject matter experts were consulted for additional studies. The detailed search strategy is available on The Community Guide website.²³

A total of 1,745 papers were screened, yielding 9 studies^24–32 for inclusion (Appendix Figure 1). Three papers were consulted for additional information on the included studies, 2 studies^33,34 related to 1 primary study²⁸ and 1 study³⁵ related to another primary study.²⁵

Table 1 provides an overview of the studies. Three studies were from the U.S.,^28,29,31 and all 3 evaluated projects within the SRTS program. Of the 6 studies outside the U.S., 2 were from the United Kingdom (UK),^24,32 3 from Australia,^25–27 and 1 from Canada.³⁰ Two studies^26,27 were purely education and promotion interventions with no infrastructure and the remaining ranged across heavily infrastructure,^24,29,32 a mix of infrastructure and promotion or education,^28,31 and mostly promotion or education with small infrastructure.^25,30

Table 1 provides additional details regarding the projects, schools, and students that were targeted. The number of projects and schools included in the U.S. studies of SRTS interventions were: 48 projects involving 53 schools in the national study,²⁸ 125 projects involving 350 schools in the California study,³¹ and 124 schools in the New York City study.²⁹ The Canadian study³⁰ involved 13 schools and the 2 UK studies^24,32 evaluated a total of 12 different projects but did not report the number of impacted schools. Most of the interventions were for elementary or primary school populations. Hence, the number of interventions evaluated, from an evidence perspective, constitutes a much larger number than a simple count of the included studies. The U.S. national study of the SRTS program²⁸ reported a median student body of 675 per participating school and the study of SRTS in California³¹ reported that 53% of projects undertaken were associated with student populations in excess of 1,000. Table 1 shows that the majority of studies reported the change in travel modes due to intervention, in particular the increase in travel by walking or bicycling following the intervention. The cost of intervention was reported by all 9 studies. Two studies in the U.S.^29,31 and 4 studies outside the U.S.^24,25,30,32 etimated benefit–cost ratios, and 2 studies from Australia^26,27 estimated cost per disability-adjusted life year averted.

The cost of the intervention from the 9 studies are provided in Table 2, along with components included in the estimate and the quality of the estimate. Cost per school or cost per project is shown, wherever possible. Two^27,30 of the estimates for intervention cost were of good quality and 7 studies^{24–26,28,29,31,32} were of fair quality. The most frequent reasons for assignment of quality limitations were: reporting funded amount without details by components or matched funding from local sources, failure to include cost of volunteer and in-kind contributions, and failure to include infrastructure component in some studies and non-infrastructure in other studies.

The grand mean of cost per school from the 3 U.S. SRTS studies was $152,243. The mean cost per school was similar for the 48 projects (53 schools) in California³¹ and for the 125 projects (350 schools) in multiple states,²⁸ at $186,576 and $179,012, respectively. On the other hand, the SRTS program in New York City²⁹ cost $91,140 per school. The difference in cost may be due to the relatively less infrastucture-heavy components in the New York City projects, which primarily improved sidewalks and crossing areas.²⁹ By contrast, the multistate study²⁸ and the California study³¹ evaluated projects that included some or all of the following in intervention cost: sidewalk construction or improvement, crosswalks, traffic calming measures, and bicycle paths and facilities. Projects in the UK had even greater infrastructure components than the U.S. SRTS projects, which may account for their higher cost of $226,753²⁴ and $664,864³² per project.

Table 3 provides the quality assessment of the estimates for benefits reported by 8 studies.^{24–27,29–32} The estimates are not presented in Table 3 because the basis of the estimates differed widely in both time horizon and in geographic scope; instead, the estimates and methods behind them are described in the Cost Benefit section and in Table 4. There were 4 good quality estimates for benefits^24,25,30,32 and 2 that were fair quality.^29,31 The most frequent reasons for assignment of quality limitations were: benefits based only on 1 impact such as injuries or fatalities, long time horizon of 30 or 50 years, short time horizon of 1 year, ATS change based on self-report or counts of users observed on routes, and ATS change included adults. Two estimates of cost per disability-adjusted life years averted from 2 studies^26,27 were assigned limited quality because they accounted for benefits from from averted obesity only, and was considered a fatal flaw for the present review. These 2 limited quality estimates were excluded from further consideration.

Estimates along with assessed quality for cost benefit and its component parts are shown in Table 4 from 2 U.S. studies^29,31 and from 4 non-U.S.^24,25,30,32 studies. One estimate³⁰ is rated as good for cost-benefit and the remaining estimates are all of fair quality. Table 3 also shows the sources and methods used to estimate the intervention cost and economic benefit, along with the geographic area and time horizon. The median benefit to cost ratio reported by the 6 studies was 5.8:1 (IQR=3.9:1–9.1:1). The study of the SRTS program in California³¹ reported a benefit–cost ratio of 0.74 over a very short 1-year time horizon and the study of the SRTS program in New York City²⁹ reported a benefit–cost ratio of 22.1:1 over a very long 50-year time horizon. Available information allowed the present reviewers to re-compute the benefit to cost ratios based on a 2-year time horizon for these 2 studies. For the recomputed estimates, the median benefit to cost ratio from the 6 studies was 4.4:1 (IQR=2.2:1–6.0:1). The median benefit to cost ratio for the infrastructure-heavy projects from the U.S.^29,31 and those from the UK^24,32 was 3.5:1 (IQR=1.7:1–6.4).

The number of people who can reasonably choose an active mode of travel to school and the proportion that actually did so at baseline and post intervention are needed to evaluate the effectiveness of ATS interventions.The omission by U.S. studies of other health and environmental benefits from ATS interventions substantially understates the plausible total economic benefits. Separate estimates for the components of economic benefits from ATS interventions should be reported. It would be useful from the perspective of policymakers from different government agencies to know what the contribution to total benefits were from: traffic injuries/fatalities, pollution, traffic gridlock, public safety and crime, physical activity, overweight and obesity, and academics and learning. Some components may have greater significance to their mission and objectives than others.