Interventions to Promote Healthy Meals in Full-Service Restaurants and Canteens: A Systematic Review and Meta-Analysis

Out-of-home eating is increasing, but evidence about its healthiness is limited. The present systematic review and meta-analysis aimed to elucidate the effectiveness of full-service restaurant and canteen-based interventions in increasing the dietary intake, food availability, and food purchase of healthy meals. Studies from 2000–2020 were searched in Medline, Scopus, and Cochrane Library using the PRISMA checklist. A total of 35 randomized controlled trials (RCTs) and 6 non-RCTs were included in the systematic review and analyzed by outcome, intervention strategies, and settings (school, community, workplace). The meta-analysis included 16 RCTs (excluding non-RCTs for higher quality). For dietary intake, the included RCTs increased healthy foods (+0.20 servings/day; 0.12 to 0.29; p < 0.001) and decreased fat intake (−9.90 g/day; −12.61 to −7.19; p < 0.001), favoring the intervention group. For food availability, intervention schools reduced the risk of offering unhealthy menu items by 47% (RR 0.53; 0.34 to 0.85; p = 0.008). For food purchases, a systematic review showed that interventions could be partially effective in improving healthy foods. Lastly, restaurant- and canteen-based interventions improved the dietary intake of healthy foods, reduced fat intake, and increased the availability of healthy menus, mainly in schools. Higher-quality RCTs are needed to strengthen the results. Moreover, from our results, intervention strategy recommendations are provided.


Introduction
The change in modern living due to urbanization and globalization [1] and the lack of sufficient free time to dedicate to home cooking have increased families' consumption of daily meals out of the home [2]. Restaurants, schools, workplace canteens and food stores providing prepared meals are the preferred food services by both children and adult populations [3,4].
Consequently, eating out of home is associated with a unhealthy diet [5] due to the lower consumption of fruits and vegetables [6]. Furthermore, comparisons of the nutritional profile of foods have shown that meals prepared out of the home are higher in energy density, fat and sodium and lower in calcium and fiber than foods prepared at home [7]. Thus, consumers of out-of-home meals may report important long-term health implications, such as obesity [8] and related chronic diseases [9]. In this regard, people are paying more attention to the healthiness of food when eating out of home [10], demanding

Screening
Initial screening of the title, abstract, keywords and publication type was conducted by two reviewers independently (F.M.; L.T.). Full-text screening of potentially relevant studies was independently performed by the same two reviewers (F.M.; L.T.) based on the inclusion and exclusion criteria, and disagreements were resolved by a third reviewer (E.L.). Final doubts about the eligibility of a particular study were resolved through discussion between the three reviewers for further confirmation and consensus.

Inclusion and Exclusion Criteria
The inclusion criteria used for the selection of eligible articles in this review were (a) controlled trials, with or without random assignment, published from 2000 to 2020 to focus the search on healthy eating interventions in full-service restaurants and canteens, conducted in contemporary circumstances; (b) English, Spanish or Italian language articles; (c) articles describing full-service restaurant and canteen-based interventions aimed at improving menu offerings and increasing the offerings and demand for healthier meals as the primary or secondary outcome; (d) trials that included a control group (CG) that did not receive the intervention; and (e) trials that presented both pre-and postintervention measurements of the intervention group (IG) and the CG and the p-values of the difference between groups.
Articles were excluded when (1) they did not fulfill the abovementioned criteria; (2) they used the pretest condition as the CG; or (3) the authors of the article were not able to give further details about the intervention results when personally asked by the authors of the present paper.
If necessary, further information about the results was collected by emailing the corresponding authors [25][26][27][28][29], especially when it was not possible to deduce the information directly from tables and figures.
The extracted results included mean changes from baseline to postintervention or follow-up and significant differences between groups in changes from pre-to postintervention. For each variable examined in between-group comparisons, differences were considered significant at p-values ≤0.05.

Data Synthesis
For a better evaluation of intervention effectiveness, the included interventions were divided according to the following: (1) outcome category (dietary intake, food availability and food purchase); (2) strategies applied (consumer-and/or establishment-based); and (3) intervention setting (school, community, and workplace) reflecting the age of the target population, i.e., children and/or adults.
Moreover, the included interventions were classified similarly to previous studies as follows [30,31]: (1) effective, when all the measured variables indicated a statistically significant change from baseline to post assessment in favor of the IG compared to the CG; (2) partially effective, when some variables included in the study changed significantly favoring the IG and any variable changed favoring the CG; and (3) not effective, when any significant changes occurred or when a change favoring the CG occurred. For the interpretation of the final effectiveness of the systematic review, an intervention was considered effective when the corresponding study reported it to be totally and/or partially effective.

Outcomes
The included studies focused on different outcomes, which were grouped into three major categories, as described previously. Specifically, (1) the dietary intake outcome category referred to the increase in the study population's consumption of healthier meals, which, according to the World Health Organization (WHO) recommendations, requires the consumption of more fruits and vegetables, the limitation of the consumption of saturated and trans fats and sugar and salt, and a balanced energy intake [32]; (2) the food availability outcome category referred to the change in the offerings of healthy and/or unhealthy food items (in terms of quality and quantity) in restaurants and canteens, which represents one of the highest-impact interventions for changing the population's dietary behavior [33]; and (3) the food purchase outcome category referred to the change in consumers' food selection towards the selection of healthier food options offered in restaurants and canteens, which is directly related to the increase in the availability of such options to satisfy consumers' demands [34].

Data Analysis
The meta-analysis was performed with Review Manager 5.4.1 and STATA 16.1 (Stat-aCorp. 2019. Stata Statistical Software: Release 16. StataCorp LLC, College Station, TX, USA) when at least three of the included intervention studies presented similar outcome variables and units of measure. Meta-analysis was performed including both RCTs and non-RCTs, and then it was repeated by excluding non-RCTs to assure higher quality results. Studies were analyzed with a random effect model when the heterogeneity of the studies was evaluated over 75% by the I 2 statistic, with the results expressed as odds ratios (ORs). When the heterogeneity was <75%, the fixed effects model was used, and the results were expressed as the risk ratio (RR) [35]. Intervention studies that presented the same measurement units and outcomes were analyzed in subgroups of studies. If the SD, SE or 95% CI values were not available in the original articles, the intervention studies were not included in the meta-analysis. A p-value of ≤0.05 was considered statistically significant.

Risk of Bias and Quality Criteria
The risk of bias and quality assessment of the included intervention studies was performed using the standardized framework of the Quality Assessment Tool for Quantitative Studies Dictionary developed for the Effective Public Health Practice Project [36]. Each included intervention study was evaluated as weak, moderate or strong for six of the eight specific categories: selection bias, study design, confounders, blinding, data collection methods, and withdrawals and dropouts. Then, the overall quality of the studies was appraised based on a 3-point rating scale including strong (no weak ratings), moderate (one weak rating) or weak (two or more weak ratings).

Results of the Search
A total of 8537 articles were retrieved from the search of the Medline, Cochrane Library and Scopus databases ( Figure 1). A total of 731 duplicates were removed, resulting in 7806 articles for title and abstract screening. Of these, 7653 were excluded because they were irrelevant for the present review by title and abstract screening. The remaining 153 articles were selected for further full-text screening according to the inclusion criteria. Following the screening, an additional 114 articles were excluded for not fulfilling the inclusion criteria. A total of 39 English-language articles were evaluated as eligible for inclusion, together with 2 articles resulting from cross-reference searching, for a total of 41 articles finally included in the present systematic review. The detailed general characteristics of the included studies are shown in Table 2, and the results on the mean pre-post intervention changes in the IG and CG are shown in Table S2.   Dietary intake x(IG-A, IG-C);x (IG-B).  Food purchase x ↓ Chips (%, n), ↓ chocolate (%, n), ↓ soft drinks (%, n).
The included studies in the present systematic review are sorted in the following table by RCTs and non-RCTs and by alphabetical order. N/A: not available; F&V: fruit and vegetable; WG: whole grain; RG: refined grain. : effective; x: not effective; x: partially effective. 1 : duration in weeks (-w), months (-m) or years (-y).

Samples of the Included Studies
The total sample size of the 41 included studies was 35,638 participants (IG: 18,988; CG: 16,650) ( Table 2). In particular, there were 16,824 participants in dietary intake interventions (children, school chefs and employees), 9361 participants in food availability interventions (children, school chefs, customers and club members), and 20,019 participants in food purchase interventions (children, employees, club members and customers). The study samples were varied and stratified in terms of sample size (from 28 to 3908 people) and age (children and adults, as reflected by the different settings).

Intervention Type
The intervention type was based on the strategies used. Each intervention applied different consumer-based and establishment-based strategies to achieve the evaluated outcome (Table 3). In particular, 3 consumer-based strategies were used to provide support, information and education (defined as a, b, and c) to consumers to improve their healthy food choices. Nine establishment-based strategies (defined as d to l) were applied for the improvement of the nutrition environment, including implementing menus offering healthier options and increasing the knowledge of restaurants and food service staff about healthy nutrition. Based on the strategies used in effective interventions, strategy recommendations were derived according to the outcome and setting applied (Table 4). x e, f, g x a    x a  Table 4. Table 4. Strategy recommendations derived from effective interventions included in the systematic review.

Setting Outcome Categories
Food Availability Dietary Intake Food Purchase

School
The involvement of the students' families, as a consumer-based strategy, together with the application of multiple establishment-based strategies, seemed to be effective in improving food availability in the school setting.
The application of consumer-based strategies together with the implementation of a menu with healthier options and limitation of the unhealthier ones, applied alone or in combination with other establishment-based strategies, seemed to be effective in improving dietary intake in the school setting. On the other hand, the provision of monetary incentives/rewards/recognition for the participating school canteen was not effective.
The application of consumer-based strategies together with the implementation of a menu with healthier options and limitation of the unhealthier ones, applied alone or in combination with other establishment-based strategies, seemed to be effective in improving food purchases in the school setting.

Community
No recommendation can be provided about both consumer-and establishment-based strategies.
The application of consumer-based strategies, together with establishment-based strategies such as the provision of monetary incentives/rewards/recognition for the participating restaurant or canteen, seemed to be effective in improving dietary intake in the community setting.
The application of multiple establishment-based strategies, including monetary incentives/rewards/recognition for the participating restaurant or canteen, seemed to be effective in improving food purchases in the community setting.

Workplace
Outcome not evaluated.
The application of consumer-based strategies together with the implementation of a menu with healthier options and limitation of the unhealthier ones, as an establishment-based strategy, seemed to be effective in improving dietary intake in the workplace setting; however more evidence is needed.
No recommendation can be provided about both consumer-and establishment-based strategies.
These recommendations are based on the interventions included in the present systematic review, as shown in Table 3.

Food Availability Outcome Category
A total of 3 RCTs of the 14 interventions that targeted food availability outcome [41,46,47] presented results of the analysis of menu offerings in school canteens (no "red" or banned food items and >50% "green" food items). The other 11 interventions (10 RCTs and 1 non-RCT) [37][38][39][40][42][43][44][45]48,62,72] presented food availability results in relation to the increase or decrease in healthy items (fruits and vegetables, unsweetened beverages, water, whole grains, etc.) or unhealthy items (high-fat, high-energy, high-sugar and high-sodium foods) on the menus of restaurants and canteens, and 3 of these studies also evaluated changes in the availability of nutrients [42,44,48] (Table S2).
Among these 14 interventions focused on food availability, 3 interventions effectively improved food availability for all the measured variables by increasing menu offerings of healthy food and beverage items and decreasing the offerings of unhealthy ones in the IG compared to the CG [41,43,46].
On the other hand, 7 interventions were partially effective by significantly changing the availability of only some of the evaluated variables, which were healthy/unhealthy food items offered on the menu, in favor of the IG [38,39,42,45,47,48,72]. Furthermore, 4 intervention studies reported no positive changes for any of the evaluated variables or reported negative changes for some variables in favor of the IG [37,40,44,62] (Table 2).
The 3 food availability interventions that were totally effective [41,43,46] were conducted in school settings, namely primary and secondary schools, and lasted from 6 to 14 months. On the other hand, among the 7 partially effective interventions, 1 was implemented in the community and 6 were implemented in schools, and they lasted from 1 week to 3 years [38,39,42,45,47,48,72].
Among these 16 interventions focused on food purchases, 3 were totally effective: 2 interventions effectively improved the population's purchase of healthy food items and beverages [45,66], and the other 3-arm intervention reported an increase in "green" food items purchased in only one of the IGs [28]. Another 9 interventions were partially effective in changing the population's food purchase of some of the evaluated healthy or unhealthy menu items, also according to their nutrient content (sodium, sugar, energy) [25,27,29,46,56,60,61,65,70], and the other 4 interventions reported no effectiveness for any of the evaluated variables [26,40,69,71] (Table 2).
Between the 3 totally effective interventions, 1 was implemented in a workplace setting, namely, hospital canteens [28], 1 in the school [66] and 1 a community setting, namely, sporting club canteens [45]; they lasted from 5 weeks to 2.5 years.

Dietary Intake Meta-Analysis
For the dietary intake outcome, the included intervention studies (RCTs and non-RCTs) were effective in increasing +0.24 servings/day of healthy food groups in favor of the IG (95% CI, 0.16 to 0.32; p < 0.001; Figure 2), including fruit, vegetables, whole grains, lean meat, and alternatives (poultry, fish, eggs, tofu, seeds, and legumes), dairy food items and alternatives (milk, yogurt, cheese). Specifically, +0.60 servings/day of whole grain (95% CI, 0.30 to 0.90; p < 0.001; Figure 2) and +0.21 servings/day of dairy food items and alternatives (95% CI, 0.01 to 0.40; p = 0.04; Figure 2) significantly increased in favor of the IG. Moreover, when non-RCTs were excluded from the meta-analysis, the effectiveness was also confirmed ( Figure S1). An increase of +0.50 g/day of fiber was also observed in favor of the IG for the analyzed intervention studies (95% CI, 0.08 to 0.92; p = 0.02; Figure S1). However, when non-RCTs were excluded from the meta-analysis, the effectiveness was not confirmed ( Figure S1). Furthermore, a positive decrease of −4.17 g/day of nutrients such as saturated fat, fat and added sugar (95% CI, −5.43 to −2.92; p < 0.001; Figure 3) occurred, favoring the IG. Specifically, −4.64 g/day saturated fat (95% CI, −7.21 to −2.08; p < 0.001; Figure 3) and −8.95 g/day fat (95% CI, −14.56 to −3.34; p = 0.002; Figure 3) significantly decreased in favor of the IG. However, when non-RCTs were excluded from the meta-analysis, only fat intake could be assessed since at least 3 studies were included and the effectiveness was confirmed ( Figure S1).

Discussion
The present systematic review included 41 interventions, 35 RCTs and 6 non-RCTs, and of these, 16 RCTs and 3 non-RCTs were included in the meta-analysis. Eligible interventions were full-service restaurants and canteen-based interventions aimed at increasing dietary intake, food availability, and food purchases in different settings, such as schools, workplaces, and communities. The results from the present systematic review showed that restaurant-and canteen-based interventions are effective in improving healthy dietary intake and food availability, mainly in the school setting, with a beneficial impact on children. However, there is partial evidence for the improvement of food On the other hand, no effectiveness was observed in the overall effect size for the intervention studies aimed at reducing the percentage of caloric intake derived from fat (%E/day) (dietary intake, −3.50; 95% CI, −7.24 to 0.24; p = 0.07; Figure S1). Moreover, these results were confirmed when excluding non-RCTs from the meta-analysis ( Figure S1).
Furthermore, a significant increase in the daily total caloric intake of +25.59 kcal/day (95% CI, 10.80 to 40.37; p < 0.001; Figure S1) was observed in favor of the CG and remained significant in the CG when non-RCTs were excluded from the analysis ( Figure S1).

Food Availability Meta-Analysis
Regarding the food availability outcome, the included interventions effectively reduced the risk of intervention schools offering unhealthy items on canteen menus by 47%, labeled red or banned food items and beverages (RR 0.53; 95% CI, 0.34 to 0.85; p = 0.008; I 2 = 43%; Figure 4).

Discussion
The present systematic review included 41 interventions, 35 RCTs and 6 non-RCTs, and of these, 16 RCTs and 3 non-RCTs were included in the meta-analysis. Eligible interventions were full-service restaurants and canteen-based interventions aimed at increasing dietary intake, food availability, and food purchases in different settings, such as schools, workplaces, and communities. The results from the present systematic review showed that restaurant-and canteen-based interventions are effective in improving healthy dietary intake and food availability, mainly in the school setting, with a beneficial impact on children. However, there is partial evidence for the improvement of food
Since all the studies included in the systematic review had weak quality, the metaanalysis was performed considering RCT and non-RCT intervention studies together, and it was repeated by excluding non-RCTs to assure results with higher quality.

Discussion
The present systematic review included 41 interventions, 35 RCTs and 6 non-RCTs, and of these, 16 RCTs and 3 non-RCTs were included in the meta-analysis. Eligible interventions were full-service restaurants and canteen-based interventions aimed at increasing dietary intake, food availability, and food purchases in different settings, such as schools, workplaces, and communities. The results from the present systematic review showed that restaurant-and canteen-based interventions are effective in improving healthy dietary intake and food availability, mainly in the school setting, with a beneficial impact on children. However, there is partial evidence for the improvement of food purchases, and more evidence is needed about workplaces and community settings as full-service restaurants. Moreover, when the meta-analysis was performed without considering non-RCT studies, the results were confirmed in dietary intake for increasing healthy food intake and in the reduction of fat intake.
The results are discussed considering systematic review and meta-analysis outcomes because meta-analysis contributes to evaluating the effectiveness of this type of intervention, and systematic review allows us to review the characteristics of interventions with effective results.
The included interventions in the meta-analysis demonstrated effectiveness in increasing the intake of healthy food items (whole grains, dairy products and alternatives) and nutrients such as fiber [27,43,[50][51][52]55,56,58,59,63,64,67,69] mainly in children, demonstrating that schools are a favorable environment for the promotion of healthy dietary intake. Furthermore, an increase in daily caloric intake occurred in favor of the CG [44,51,55,56,63,64,67], and effectiveness was observed for decreasing the consumption of other nutrients such as saturated fat and fat in the IG [49][50][51]64,67]. For food availability outcome, the intervention studies included in the meta-analysis were also demonstrated to be effective in reducing the risk, for the intervention schools, of offering unhealthy foods and beverages on canteen menus [41,46,47].
For interventions in the dietary intake outcome category, the present results showed effectiveness mainly in school settings, which was the preferred setting for interventions targeting these outcomes. When targeting children, an important factor to be considered in nutrition interventions is food presentation in terms of color and smell, which should be appetized to trigger food selection and consumption. Thus, repeated exposure to healthier foods presented in attractive ways could help children become more accustomed to and consume it [73]. Focusing on adults, changing dietary habits to achieve a healthier lifestyle is made more difficult by the perceived barriers, such as: lack of cooking skills and willpower; time scarcity; the need to give up one's favorite foods [74]; and social, cultural and economic conditions [75]. However, although the evidence about workplace settings is very limited in the present review, workplace interventions have the potential to change consumers' dietary behavior through the working lifespan [76]. Long-term workplace interventions for approximately one year evidenced an improvement in dietary change among the participants [77], while the included studies in this systematic review lasted less than one year. However, it is important to highlight that published evidence and its quality in workplace programs are suboptimal; thus, this conclusion needs to be verified with high-quality interventions [77].
From the present results, regarding the intervention strategies applied to improve dietary intake, the implementation of establishment-based interventions is different in the three evaluated settings. Specifically, the strategies that showed higher effectiveness in schools were the addition of healthier menu options combined with on-site support, training for the school canteen staff, performance monitoring and feedback reports (Table 4). However, in the community setting, including after school programs and recreation centers, the provision of monetary incentives, rewards, and recognition for the participating food service are effective, while these methodologies are ineffective in schools.
According to the interventions in the food availability outcome category, none of them were set in workplaces, and little evidence resulted in the community setting [45], whereas effectiveness was reported in the school setting [38,39,[41][42][43][45][46][47][48]72]. In schools, regarding the intervention strategies applied for food availability outcomes, the involvement of the participants' families, namely students and their parents in school-based interventions, through invitations to meetings, activities and the distribution of information letters, was the most effective consumer-based strategy [41,46,72]. Similarly, in a recent review focusing on family-based interventions to improve children's diets, the family involvement strategy through the provision of information, advice and monitoring was also reported to be effective in improving the food environment of school canteens, demonstrating that parents are an important component when children are targeted [78].
Children's improvements in food availability are important because their adherence persists in adulthood, whereas unhealthy food availability reinforces children's preference for nutrient-poor and ultra-processed foods [79]. The increase in healthy food availability in school settings is directly correlated with healthy food purchases, with the final aim of changing children's dietary intake [80].
On the other hand, the implementation of healthier food availability in the community setting is more difficult due to the barriers stakeholders encounter, such as the lack of demand by customers and the increased cost associated with healthy fresh foods with a short shelf life [81][82][83], but financial support and resources such as guidelines and training from established associations could help achieve such improvements [81]. Thus, future interventions aimed at increasing the availability of healthier food options in community settings should also target an increase in consumers' demands for healthy meals, as well as assure food services of the low risk of changes in their profits [84].
For the interventions in the food purchase outcome category, partial effectiveness was reported mainly in schools through the implementation of multiple consumerand establishment-based strategies, including the involvement of participants' families [25,27,29,46,60,65,66]; thus, family certainly has a good influence on children's food selection [85].
On the other hand, little evidence about effective strategies in community and workplace settings was apparent in the present systematic review; however, in community settings such as restaurants and food stores, the provision of information and communication to consumers may not be enough to achieve behavior changes such as the selection of healthier food options [26,40], whereas multiple strategies targeting changes in the food environment could be fundamental for improving customers' food purchases [45].
Moreover, effective consumer-and establishment-based strategies were derived from the included interventions to develop methodological recommendations, by outcome and setting, for the implementation of future restaurant and canteen-based interventions ( Table 4). There were some limitations in the present systematic review and meta-analysis. First, the lack of randomized controlled studies in workplace and community settings, such as full-service restaurants, limited the evidence about the adult population and the evaluation of the interventions' effectiveness. Second, the exclusion of fast-foods and chain restaurants in this systematic review and meta-analysis limited the generalizability of the results to other out-of-home settings, but it allowed us to provide specific recommendations for full-service restaurants and canteens. Third, the lack of enough evidence for the different community settings included, such as after-school programs, restaurants, sporting clubs, and recreation centers, made it difficult to detect differences in intervention strategies. Fourth, none of the included studies were set in low-income countries because of the intervention gap in the literature about middle-and low-income countries [86], limiting the inclusivity of a wider target population. Fifth, in the meta-analysis, the wide heterogeneity of the included studies in terms of outcomes and units of measure, and the huge quantity of different outcomes included, as well as the lack of specific numerical data in the articles, made it difficult to compare interventions and reduced the interventions included. Finally, the quality of most of the included studies was assessed to be of weak quality since the majority had no blinding, poor data collection methods, selection bias or confounders.

Conclusions
In conclusion, restaurant-and canteen-based interventions demonstrated effectiveness in the improvement of healthy food intake and in the reduction of fat intake and in increasing healthy menu availability, mainly in school settings. For food purchases, a systematic review showed that interventions could be partially effective in improving healthy foods. However, higher-quality RCTs are needed to strengthen the results. Moreover, intervention strategy recommendations were provided for each outcome assessed to increase the effectiveness of restaurant-based interventions implemented.
Author Contributions: Each author has made substantial contributions to the conception or design of the work (F.M., E.L., L.T., R.M.V., R.S.); the acquisition, analysis, or interpretation of data (F.M., E.L., L.T., R.M.V., R.S.); the creation of new software used in the work or has drafted the work or substantively revised it (F.M., E.L., L.T., R.M.V., R.S.). Each author has approved the submitted version (and a version substantially edited by journal staff that involves the author's contribution to the study) and agrees to be personally accountable for the author's own contributions and for ensuring that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and documented in the literature. All authors have read and agreed to the published version of the manuscript.
Funding: This publication has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No. 713679. This publication has been possible with the support of the Universitat Rovira i Virgili (URV) and Banco Santander.