Community-Based Interventions in Prepared-Food Sources: A Systematic Review

Introduction Food purchased from prepared-food sources has become a major part of the American diet and is linked to increased rates of chronic disease. Many interventions targeting prepared-food sources have been initiated with the goal of promoting healthful options. The objective of this study was to provide a systematic review of interventions in prepared-food sources in community settings. Methods We used PubMed and Google Scholar and identified 13 interventions that met these criteria: 1) focused on prepared-food sources in public community settings, 2) used an impact evaluation, 3) had written documentation, and 4) took place after 1990. We conducted interviews with intervention staff to obtain additional information. Reviewers extracted and reported data in table format to ensure comparability. Results Interventions mostly targeted an urban population, predominantly white, in a range of income levels. The most common framework used was social marketing theory. Most interventions used a nonexperimental design. All made use of signage and menu labeling to promote healthful food options. Several promoted more healthful cooking methods; only one introduced new healthful menu options. Levels of feasibility and sustainability were high; sales results showed increased purchasing of healthful options. Measures among consumers were limited but in many cases showed improved awareness and frequency of purchase of promoted foods. Conclusion Interventions in prepared-food sources show initial promising results at the store level. Future studies should focus on improved study designs, expanding intervention strategies beyond signage and assessing impact among consumers.


Introduction
Obesity is a multifactorial disease that has many contributing factors, including the food environment. Food environments in which energy-dense prepared foods (ready-to-eat foods that can be eaten outside the home or brought back or delivered to the home to eat) are readily available are associated with the increasing prevalence of obesity (1)(2)(3). As a consequence, environmental interventions may be more cost-effective, sustainable, and successful at reaching a large population than individual interventions (4). Today, Americans spend about half of their food dollars eating out (5,6) compared with 25% in 1955 (7). Americans are expected to spend $660.5 billion in 2013 on eating out, a 15-fold increase from 1970 (adjusted to current dollars) (7). The total energy intake per capita increased on average by 570 kcal between 1977-78 and 2003-06; this increase has been attributed to greater portion sizes and a greater frequency of eating out (8). Although prepared foods are generally more costly than foods prepared at home, people who have low incomes consume them at high rates (9)(10)(11). Prepared-food sources may be an important venue for efforts to reduce obesity and risks for other chronic diseases. In the past several decades, multiple interventions were conducted in venues that provide prepared foods. One review (12) showed that interventions had success in private prepared-food-source environments, such as worksite and college cafeterias. However, no systematic reviews examined prepared-food source interventions in public community settings such as carryout, fast-food, and sit-down restaurants, even though preparedfood sources are more numerous in public settings than in private settings, and low-income individuals may have little

Study selection
We initially identified 35 potential interventions. All identified food-source interventions were reviewed for inclusion according to the following criteria: 1) a focus on prepared-food sources in public community settings such as carryout, fast-food, and sit-down restaurants (although other intervention components could be included as part of the intervention); 2) a completed impact evaluation (eg, pre-post assessment, use of a comparison group, exposure assessment with or without a comparison group); 3) some form of written documentation (eg, peer-reviewed journal article, newsletter article, other published article, policy brief or report, published trial material, a project's own website or conference presentation) that included a description of the implemented intervention and evaluation findings; and 4) an intervention start date after 1990. Of the original 35 interventions, 19 met these inclusion criteria.
For 4 of the 19 interventions, only website information was available, and because we were unable to obtain further information, these 4 were also excluded from our review. Of the 15 remaining, 2 interventions lacked information on study findings and were therefore excluded. These exclusions left 13 interventions for review, data extraction, and analysis.

Data extraction
The 2 primary reviewers (S.L.K., B.B.) independently extracted and analyzed data by reviewing all documents. The secondary reviewer (J.G.) developed the system for extracting data and coding variables. The secondary reviewer also resolved discrepancies noted by the 2 primary reviewers and identified and adjudicated other discrepancies that might affect reliability and analysis (Appendix).
Primary reviewers were instructed to extract data for each variable and to organize data using the intervention as the unit of analysis. We attempted to contact a representative of each intervention to obtain more information. Seven of 13 intervention managers or lead researchers participated in semistructured telephone interviews or e-mail communications (or both) designed to gather additional information needed to complete the data tables and to resolve any inconsistencies. One intervention (Baltimore Healthy Carryouts) was implemented by our team. After the interventions were reviewed, we summarized data in 3 tables: background and intervention approach, evaluation methods used, and study findings. Within each table, we grouped restaurants into 4 categories according to similar characteristics: specialty restaurants, chain restaurants, small local restaurants, and mixed types of restaurants (both local and chain) that did not share characteristics with other restaurants or whose characteristics were not well defined. The primary reviewers organized the information about each study into these tables. The secondary reviewer confirmed data accuracy by using initial review findings, e-mail correspondence, interview transcripts, and extraction and reporting guidelines.
We implemented previously used techniques (14) for analytic assessment of the selected interventions, such as identifying a standard set of quality criteria (eg, randomization and use of control groups) and reporting on the impact among prepared-food sources and consumers. Because of the heterogeneity of outcome data, which did not permit us to calculate summary estimates of impact, meta-analytic techniques were not used.

Target populations
Most interventions (n = 9) took place in urban settings; 3 interventions focused on mixed urban and rural populations, and one focused on rural settings. Interventions were conducted in various regions of the United States: 4 in the Northeast, 4 in the West, 2 in the Midwest, and 1 in the Southeast. Only 2 interventions took place outside the United States (Table 1a, 1b, 1c, 1d). The interventions targeted a range of consumer income levels: 3 interventions (Baltimore Healthy Carryouts [15][16][17][18][19][20][21][22][23], Steps to a Healthier Salinas [25,26], and Coeur en Santé [28,29] worked in low-income areas, Horgen and Brownell (27) worked in middle-to high-income areas, and the rest did not identify the income level of the targeted population. The race/ethnicity of consumers in the interventions was generally mixed: 5 populations were predominantly white; 1, Hispanic; 1, African American; 1, Korean; and 1, mixed; 4 interventions did not specify the race/ethnicity of the targeted population.

Intervention goals
Overall, the 13 interventions shared similar goals. All sought to promote more healthful menu items at prepared-food sources with the associated goal of increasing sales of these foods. Some interventions had additional goals. The larger goal of Shape Up Somerville (32-36) was to reduce childhood obesity, and their restaurant intervention (33) was viewed as an environmental component of that work. The additional goal of Baltimore Healthy Carryouts (15)(16)(17)(18)(19)(20)(21)(22)(23) was to assess the feasibility of a carryout intervention. Other projects stated that they sought to achieve their overall goals by specific means, such as menu labeling (Tandon et al [31] and Smart Menu program [37,38]) and price changes (Horgen and Brownell [27]).

Community promotion
Four interventions did not promote their programs outside the prepared-food sources or in the general community. Of the 7 interventions that promoted their programs widely, a variety of approaches were used, including newspaper advertisements (n = 5), promotion at community events (n = 3), and leaflets and newsletters (n = 3). These promotions were intended to increase awareness of the program and to direct consumers to prepared-food sources participating in the intervention.

Evaluation methods used and key findings
Feasibility and process evaluation of interventions Most interventions (n = 8) collected information on feasibility. In general, this information was in the form of interviews and informal discussions with owners and managers of the prepared-food sources. These same 8 interventions also conducted some form of process evaluation (usually through store visits) to assess, for example, whether signage was displayed and healthful options were available.
Overall, the level of feasibility was moderate to high for intervention implementation (Table 3a, 3b, 3c, 3d). The level of acceptability was generally high among participating food-source owners. Menu labeling was particularly acceptable among multiple interventions. However, no trial assessed program acceptability among consumers.

Impact of interventions on prepared-food sources
Seven interventions assessed the impact of the intervention on prepared-food sources. Four of these 7 interventions (Baltimore Healthy Carryouts, Healthy Restaurant, Healthy Howard, and Shape Up Somerville) interviewed food-source owners to assess whether sales of healthful options increased. Four used data from food-source registers; 2 interventions collected sales receipts (Baltimore Healthy Carryouts [15][16][17][18][19][20][21][22][23] and Horgen and Brownell [27]). Three interventions -Healthy Howard (43), Healthy Restaurant (44)(45)(46), and Shape Up Somerville (32-36) -collected information other than sales data from the owner or manager of the prepared-food source, including information on nutrition awareness.
Of the 6 interventions that reported an assessment of psychosocial factors among consumers, most (n = 4) found an increased awareness of the intervention and its goals. Of these, Coeur en Santé (29), TrEAT Yourself Well (30), and Healthy Restaurant (44)(45)(46) showed improvements in acceptability of healthful food among customers.

Discussion
This is the first systematic review of interventions in community-based prepared-food sources. Results are promising, showing that cost-effective methods (eg, labeling foods as healthful) may have a significant impact on prepared-food source sales and customer behavior. Most approaches centered on signage to promote existing healthful food choices. Several worked with kitchen staff to improve low-fat food preparation practices, and several conducted formal menu analyses to determine more healthful choices for promotion. These strategies, though resource intensive, show that it is important to engage prepared-food source staff in determining more healthful options.
Although interventions in prepared-food sources are increasingly being considered viable public health interventions, it is clear from this review that the evidence base for these interventions is deficient. Study designs tended to lack comparison groups, treatment conditions were not randomly assigned, and selection criteria were not readily apparent; these inadequacies cast doubt on the generalizability of findings.
Many of the interventions included in this review were not formal studies but rather certification or campaign programs operated by local health departments. These programs were voluntary, which may explain why they varied in levels of reach. Low reach may have been due to the actual burden placed on restaurant owners by the intervention or the lack of potential benefits perceived by them. Partnerships between local public health departments and academic institutions may help overcome some of these limitations through improved social marketing of program strategies and benefits. For example, local public health departments could initiate the intervention and manage recruitment while an academic institution partner conducts a full impact assessment, including psychosocial surveys and sales data collection.
We found no clear preference for a conceptual or behavior-change framework. A few interventions used social marketing approaches, but these interventions generally did not incorporate key social marketing components, such as formative research, audience segmentation, or targeting of messages (50). New frameworks for operationalizing prepared-food source interventions need to be developed; such frameworks should incorporate elements of theories on the food environment (51,52) and behavior change.
Although many interventions showed promising results, most lacked adequate measurement of the impact of the intervention on consumers. Of studies that examined consumer outcomes, most assessed only awareness of the health promotion campaign and food purchasing frequency. More sophisticated assessments -based on behavior-change theories -are needed to evaluate such outcomes as food purchasing patterns and dietary intake.
Almost half of the interventions lacked formal formative research -a major omission, especially considering that food environment research is an emerging area of public health programming. Formative research involves qualitative and quantitative data collection, which aids in program development (53)(54)(55)(56); many interventions in our study would have benefited from this kind of preliminary research. In addition, most interventions focused only on promoting existing healthful options at prepared-food sources. Only 1 intervention sought to introduce new, more healthful foods. Interventions need to pay more attention to actively changing the food environment in prepared-food sources by increasing the availability of healthful food options.
Another approach worth exploring is price changes. Two interventions showed positive results by reducing food prices (Baltimore Health Carryouts [15][16][17][18][19][20][21][22][23] and Horgen and Brownell [27]). Ample evidence supports the use of price reduction as a means of promoting more healthful options among consumers (57)(58)(59). Moreover, research on pricing may support the hypothesis that the increased prevalence of obesity may be attributed to greater consumption of soda and chips that have artificially low prices because of government subsidies for corn and soybeans (4,60,61). If nonhealthful and healthful food prices were comparable, consumers might be more likely to purchase healthful foods. Baltimore Healthy Carryouts demonstrated that reducing the price of healthful foods not only increased sales of healthful foods but also increased total carryout revenue (17,20).
Our findings parallel in many ways the findings of recent interventions in small retail food stores (14). Both types of interventions focused on modifying the food environment and promoting healthful choices through point-of-purchase materials. A key difference, however, was that most research on small food stores has focused on food deserts in primarily low-income racial/ethnic minority populations. Only a few of the interventions in our study took place in lowincome settings. More carefully evaluated interventions in food sources in low-income settings are needed. The high rates of obesity and other chronic diseases are related not only to the availability of healthful choices in retail food stores but also to access to healthful foods in prepared-food sources (15,(62)(63)(64). Although a recent longitudinal study (63) showed mixed results on the relationship between the availability of grocery stores and diet-related outcomes, it demonstrated that the availability of fast food was related to fast food consumption in a low-income population. Low-income populations tend to rely on local prepared-food sources because they often work more than one job and do not have time to cook at home (65). One study found that during the 2007-09 recession, the middle quintile of households (the middle 20% of the income distribution) cut spending on food away from home by 20%, whereas the lowest quintile of households cut such spending by 12%, suggesting a greater reliance on food away from home among lower quintile households (66). Improving the availability of healthful food in prepared-food sources may be an effective way to promote dietary improvement in low-income settings.
This systematic review has some limitations. Many of the interventions reviewed were implemented by health departments that lacked the resources to conduct comprehensive evaluations and publish the findings in peer-reviewed publications. Thus, we included findings from the gray literature. Because of the wide variability in measures used and in impact assessments, we were not able to develop summary estimates or compare measures or impacts directly among interventions. These limitations emphasize the need for standardization of measures used by interventions and the need for further reviews that assess different strategies (14).
Our review lays the groundwork for further exploration of strategies to increase more healthful food options in community-based prepared-food sources. Many interventions showed that changing the prepared-food environment may improve sales and awareness of more healthful foods and improve purchasing and consumption behaviors. Interventions can be strengthened through comprehensive formative research and quantitative assessments of process and impacts. With these additions, future studies will be able to assess the relative effectiveness of different strategies and create standards of practice in prepared-food sources. Tables   Table 1a.

Key Components of Interventions on Community-based Prepared-Food Sources: Interventions Conducted in Specialty Restaurants
Characteristic Baltimore Healthy Carryouts (15)(16)(17)(18)(19)(20)(21)(22)(23) Good for You (24) Steps to a Healthier Salinas (   Pilot survey to indicate customer preferences; presented to owners Includes small, locally owned "mom-and-pop" establishments that include but are not limited to take-out and sit-down restaurants and restaurants that focused on specialty foods; it excludes chain restaurants.  Includes restaurants (both local and chain) that did not share characteristics with other intervention restaurants or whose characteristics were not well defined.  Includes small, locally owned "mom-and-pop" establishments that include but are not limited to take-out and sit-down restaurants and restaurants that focused on specialty foods; it excludes chain restaurants.
Feasibility assessment measures include acceptability, operability, and perceived sustainability. Process evaluation measures include dose, reach, and fidelity, which indicate how well the program was implemented according to plan.
Consumer impact measures included psychosocial, behavioral, and health outcomes. Includes restaurants (both local and chain) that did not share characteristics with other intervention restaurants or whose characteristics were not well defined.
Feasibility assessment measures include acceptability, operability, and perceived sustainability. Process evaluation measures include dose, reach, and fidelity, which indicate how well the program was implemented

Consumer behavioral impact results
Other results

Sustainability
Policy results, implications The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors' affiliated institutions.
The RIS file format is a text file containing bibliographic citations. These files are best suited for import into bibliographic management applications such as EndNote , Reference Manager , and ProCite . A free trial download is available at each application's web site.