Analysis of state-level FSG policies included historical and contemporary state statutes, promulgated administrative regulations, and executive orders. A state statute is a codified state law. Regulations are administrative rules issued by government agencies that have the force of law because they are adopted under authority granted by the state. A state executive order is a Governor’s declaration that has the force of law and typically requires no action by the state legislature.

Authors used a multi-pronged search strategy to identify FSG policies enacted between January 1, 2015 and April 1, 2019. Like the previous study, the commercial legal research database, Westlaw Edge (Thomson Reuters, Toronto, Canada), was used as the primary source to apply search strings and identify potential FSG policies for inclusion.⁷ Two enhanced search strings² were run to ensure authors captured FSG policies dealing with healthier food procurement and heathier foods being served in state run prisons. A secondary legal database, CDC’s Policy Tracking Surveillance System,¹⁶ was searched to ensure no state FSG policies that met study inclusion criteria were missed.

The lead author and 1 legal researcher managed application of the FSG search strings in the policy research databases and made final determinations on inclusion or exclusion of policies. To meet inclusion criteria, all FSG policies had to specify the development of, or refer to, nutrition standards that apply to foods and beverages served, sold, or procured for adult populations in government-owned or -controlled facilities and programs, including where conferences and on-site or off-site events are held. Alternately, policies could specify the development of a task force or other government committee to recommend FSG standards for government-run programs. Policies were excluded if they primarily focused on individuals aged 18 or younger, or food-insecure populations utilizing the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC). These national feeding programs already follow existing nutrition standards set by federal authorities. Policies focusing on meals served to medically vulnerable populations were also excluded, such as those in nursing homes, in-patient hospitals, or rehabilitation centers. These populations have nutritional needs that differ from the general population, and application of FSG standards may not be medically appropriate.¹

Once all FSG policies meeting inclusion criteria were identified, authors began the content analysis and coding phase of the study. Two trained coders with public health law expertise, and a third coder with nutrition expertise, analyzed the full text of each policy to assess the presence or absence of key attributes using an updated FSG Tool (Appendix 1). Independent coders utilized a traditional present (1) and absent (0) schema to code policies, consistent with policy coding best practices.¹⁷ First, each of the 3 trained coders independently reviewed and classified FSG policies in 1 or more of the FSG Tool’s food service setting categories: (1) vending machines on government-owned or -leased property (“vending machines”); (2) meals on government-owned or -leased property (“meals”); (3) food or beverages made available at government events or meetings (“events or meetings”); (4) food service guideline task force (“task force”); or (5) government food procurement (“food procurement”). FSG policies impacting foods and beverages across multiple setting categories (e.g., vending machines, cafeterias, meetings) were coded as “all foods.” Next, FSG policies were coded based on how well they align to variables in 4 key FSG domains: (1) nutrition standards referenced; (2) behavioral design strategies and selection of healthier choices, (3) supports for FSG implementation; and (4) facility efficiency practices that encourage environmentally responsible practices.

Once all FSG policies were initially coded, inter-rater reliability was determined among the 2 primary policy coders (96%), with the third reviewer’s codes acting as a tie breaker in instances of discrepancy. Each policy was awarded an average score in each of the 4 FSG policy domains. This score was calculated by taking the overall proportion of FSG variables coded as present (numerator), divided by the total number possible in each domain (denominator). For example, if a policy included language consistent with only 1 variable in the behavioral design domain, the numerator would be 1 and the denominator would be the total number of behavioral design variables, or 4 in this example. The policy would be awarded a score of 25% in the behavioral design domain. Finally, the 4 FSG Tool domain scores for each FSG policy were used to calculate an overall comprehensiveness score (average of the 4 sub-domain scores). This FSG comprehensiveness score represents an adopted FSG policy’s overall strength and quality as it relates to potential public health impact.

The FSG Tool was originally developed in 2015 and was adapted from the National Cancer Institute’s validated instrument, the Classification of Laws Associated with School Students (CLASS) system, which is used to score state laws for physical education and nutrition in schools.¹⁸ A complete description of the FSG Tool and its development can be found elsewhere.⁷ In 2019, the FSG Tool underwent 2 modifications to capture more nuanced FSG policy attributes. First, the tool was modified to award minimum credit to any policy that includes language encouraging or requiring adherence to DGA nutrient requirements, even if failing to name standards operationalized for use in food service venues. Alternatively, FSG policies referencing standards that (a) align to current DGA nutrition science and (b) are tailored for use in food service venues received an additional code, thereby, earning a higher score in the FSG nutrition domain. This change allows researchers to quantify shifts over time in states awareness of published guidelines that both align to current nutrition science, DGA, and are specifically for use in food service operations.

A second modification to the FSG tool included pulling out facility efficiency and environmental sustainability policy attributes into a standalone domain score called “Facility Efficiency”. This was done to acknowledge an increasing emphasis on food service operations and the impact on environmental sustainability and the local food system. In the previous study period (2007–2014), these attributes were still calculated in the overall FSG comprehensiveness score; however, they were not easily identifiable as they were embedded in the FSG Tool’s implementation supports domain. By creating a standalone domain, authors hope to provide states with data to assess progress and educate policy makers on the interconnectedness of efficient food service operations, resilient local food systems, and human health.

Since 2007, there have been a total of 20 FSG policies adopted by 11 states and D.C. (Figure 2). Across the 2 study periods, FSG policy adoption occurred most often in the southern and western regions of the U.S., with California adopting the greatest number (n = 4) of FSG policies to date. Since 2007, state Agency or Departmental rules and regulations were the most utilized legislative mechanism for adopting FSG policies, compared to state executive orders or state statutes (data not shown).

As shown in Table 1, nutrition domain scores ranged from 80% to 100%, with 4 out of 5 FSG policies receiving a score of 100% as they explicitly reference DGA-aligned nutrition standards that are also translated for use in food service venues. The 1 policy that did not reference such standards—a California prisons meals law—still earned a nutrition score of 80% because it referenced the 2008 California Food Guide, which encourages alignment to the DGA, but fails to name standards operationalized for use in food service. As shown in Table 2, the most common set of nutrition standards referenced in adopted FSG policies were Food Service Guidelines for Federal Facilities published by CDC in 2017 (originally known as the HHS/GSA Health and Sustainability Guidelines).⁵

Behavioral design scores were the lowest among all FSG Tool domains, ranging from 0% to 25% (Table 1). Behavioral design strategies were least often included in adopted FSG policies, with California and Oregon incorporating none. The D.C., Vermont, and Louisiana FSG policies each included 1 of the 4 behavioral design domain attributes. Specifically, Louisiana and Vermont’s FSG policy required nutrition labeling on foods sold; D.C.’s policy required that healthier foods and beverages be the only options advertised in the park concessions and vending venues, including on recreational sports scoreboards.

As shown in Table 3, adopted FSG policies included zero to 4 of the FSG Tool’s 6 implementation supports: (1) naming a government body to oversee implementation, (2) routine compliance checks, (3) conduct vendor training on healthy food service, (4) review of standards after a period to ensure alignment with current nutrition science (5) requirement that a specific percentage of products meet nutrition standards, and (6) authorization of funding to support FSG implementation. Four of the 5 FSG policies named a government body to oversee implementation, but only Oregon and Vermont specifically authorized regular compliance checks to assess on-going FSG implementation (Table 3). No policies addressed vendor training on healthy food service or nutrition standards, and no state expressly dedicated funds for FSG implementation efforts. Both Washington D.C and Vermont required a percentage of offerings that must meet nutrition standards in policy language, while Oregon took a more innovative approach, and included policy provisions that financially incentivized vendors to adhere to a tiered percentage of required healthier offerings. Vermont was the only state to authorize post-implementation review of required nutrition standards to ensure alignment with current nutrition science in the DGA.

Facility efficiency scores among the 5 adopted FSG policies ranged from 0% to 100% (see Table 1). To earn credit in this domain, adopted FSG policies had to expressly require, authorize, or encourage products that are locally sourced, certified organic, or use of another environmentally friendly practice, e.g., increasing plant-based food offerings. For the first time in CDC’s tracking of adopted FSG policies, authors observed 2 states (Oregon and Vermont) that included policy provisions that explicitly require or incentivize procurement of locally sourced foods in state-run food service operations.

Tracking and analyzing state-level adoption of FSG policies can help researchers better describe the policies public health impact over time. However, there are some limitations of the present study. First, we did not verify FSG implementation post-policy adoption, or assess associated outcomes. Policy adoption is an important first step, but further research is needed to determine which supports correlate with higher levels of implementation and impact. In addition, due to the evolving nature of adopted FSG policies, authors made minor modifications to the policy search strings, inclusion criteria, and scoring criteria during this study period. This was done to ensure recent FSG policy innovations were captured, limiting authors ability to compare all FSG classification domain scores over time. Finally, it is possible that facility efficiency and food safety standards were missed because they are sometimes included in a state’s non-FSG-related policies or regulations.