Citizen science approaches to crowdsourcing food environment data: A scoping review of the literature

Globally, the adoption and implementation of policies to improve the healthiness of food environments and prevent population weight gain have been inadequate. This is partly because of the complexity associated with monitoring dynamic food environments. Crowdsourcing is a citizen science approach that can increase the extent and nature of food environment data collection by engaging citizens as sensors or volunteered computing experts. There has been no literature synthesis to guide the application of crowdsourcing to food environment monitoring. We systematically conducted a scoping review to address this gap. Forty‐two articles met our eligibility criteria. Photovoice techniques were the most employed methodological approaches (n = 25 studies), commonly used to understand overall access to healthy food. A small number of studies made purpose‐built apps to collect price or nutritional composition data and were scaled to receive large amounts of data points. Twenty‐nine studies crowdsourced food environment data by engaging priority populations (e.g., households receiving low incomes). There is growing potential to develop scalable crowdsourcing platforms to understand food environments through the eyes of everyday people. Such crowdsourced data may improve public and policy engagement with equitable food policy actions.


| INTRODUCTION
Dietary risks describe diets high in processed foods and beverages that contain excessive amounts of energy, fat, sugar, and salt and diets low in wholegrains, fruits, vegetables, legumes, nuts, seeds, and dairy.These diet patterns are leading contributors to a high body mass index (BMI) and chronic non-communicable diseases (NCDs) such as diabetes, cardiovascular diseases, and some cancers. 1 In many countries, the prevalence of NCD burdens is greater among those experiencing the most, compared with the least, socioeconomic disadvantage. 2There is consensus in the public health nutrition literature that dietary risks have become prevalent because of our food system and environment. 3Food environments are defined as the interface where people interact with the wider food system to acquire and consume foods and beverages-generally encompassing the availability, affordability, nutritional quality, labeling, and promotion of foods and beverages. 4In today's food environments, unhealthy foods and beverages are highly available, convenient, affordable, and heavily marketed, making them easy to purchase and consume in excess. 5tensive literature and calls from leading health organizations such as the World Health Organization, suggest that government-led policy actions are required to transform our food environments to protect and promote healthy foods and beverages over unhealthy options. 5,6While some jurisdictions have introduced policies to improve front-of-pack nutrition labeling and tax sugar-sweetened beverages (among other policy measures), political commitment towards adopting and implementing comprehensive food policy strategies has been inadequate. 5,7This is, in part, because of the complexity of monitoring the healthiness and associated impacts of our dynamic food environments, in a timely, low-resource and policyrelevant manner. 8Currently there are no rigorous and sustainable systems in place to monitor various aspects of our food environments, therefore little way of holding governments and industry accountable to public health objectives.Engaging citizens in this process is increasingly recognized as an important method to galvanize civil society support for policies that create healthy and equitable food environments. 9owdsourcing is one potential approach to increase the extent to which food environment data can be collected by engaging citizens as sensors or volunteer computing experts (typically aided by digital tools such as mobile applications or cameras). 10p506) Crowdsourcing can be further defined as the involvement of citizens in the building of a dataset through their active contributions to data collection and analysis.This approach offers a new way to problem solve by actively drawing on a wider range of perspectives and lived experience-which can be done on a large scale if resources are available. 11,12To date, crowdsourcing has most commonly been utilized to collect data to monitor and engage the public with biodiversity and agriculture including pest control methods and virtually participating in marine research by analyzing images. 13,14In recent years, crowdsourcing methods have been increasingly employed within various health fields, including public health. 156][17] Nevertheless, this previous review did not include studies on how crowdsourcing can be used to examine barriers and enablers to accessing healthy food or empower priority populations to advocate for healthier food environments. 18,19Additionally, we wanted to understand the uses of active crowdsourcing, in which citizens are actively engaged in the research project through either civilian mobilization or policy change within the community.To date, no systematic synthesis has been conducted to summarize the extent and nature of using active crowdsourcing as a method to monitor and engage citizens with food environments for public health.This scoping review aimed to address this research gap, thereby building the evidence base for and guiding future use of crowdsourcing in public health nutrition.

| METHODS
This article is reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines. 20Crowdsourcing within the wider area of public health and research is a new field; therefore, a scoping review was identified as the most appropriate method to understand what constitutes crowdsourcing in relation to food environments (as they are understood from a public health perspective) and summarize the implications for future work in this field. 21

| Eligibility criteria
Inclusion criteria for this review were intentionally broad to capture the objectives of our scoping review. 21Reviews, qualitative and quantitative studies were eligible for inclusion if they met the following criteria: (i) the study aim aligned with the definition of "crowdsourcing" by which people are actively engaged in collecting data (i.e., engaging citizens to actively participate and act as sensors to contribute to the targeted and systematic collection of data; aligning with the first principle of citizen science "to actively involve citizens in scientific endeavours" 22 (p1) ), (ii) the study examined topics within the scope of food environments as defined by Turner et al. 4 (i.e., food availability, affordability, labeling, promotion, and nutritional quality-including built and digital environments), and (iii) articles were in English and published in or after 2011 (based on the recency of the articles located during our preliminary searches).Articles were excluded for the following reasons: (i) the study aim did not explicitly involve crowdsourcing (e.g., the passive involvement of people in research); and (ii) the study population consisted of only participants with chronic health conditions and/or special dietary requirements (i.e., this population group may experience unique barriers to their food environments that are not generalizable).
Both AB and CC have expertise in the topic of citizen science and were regularly consulted regarding the inclusion criteria.

| Article screening and selection
A preliminary search was conducted (ALM and CZ) to establish appropriate terminology that would retrieve articles that met our inclusion criteria.Four search term categories were identified: "citizen science/ crowdsourcing," "crowdsourcing tools" (e.g., camera and mobile application), "food," and "environments" (see Appendix S1 for full Medline search strategy).These search terms and relevant synonyms were systematically applied across five electronic databases: Medline, Informit, CINAHL, Global Health, and Scopus.Subject headings were used where applicable, and the search terms were translated for each database as necessary.
After the database searches were completed, search results were exported to EndNote and then Covidence, where duplicates were removed.The remaining titles and abstracts were screened against the eligibility criteria by five researchers (JM, CZ, RC, AC, and CD).Two researchers (JM and CZ) worked collaboratively to screen the full-text articles considered to be relevant and identify all eligible studies.If it was unclear whether the methods used met the inclusion criteria, the two researchers discussed their views until consensus was reached.
Grey literature was also included in the search, including reference lists of all included articles, known reviews, and the first 200 results from Google Scholar (after which the results became increasingly irrelevant).The final search was conducted in November 2021.

| Data extraction
Data from the included articles were extracted into a Microsoft Excel template.This included the following information: author(s), title, year of publication, country, number of participants, gender, age, population characteristics (if specified), sampling methods, data collection methods, and main findings.

| Critical appraisal
The purpose of our scoping review was to map the extent and nature of the use of crowdsourcing to monitor and engage people in food environments.It was beyond the scope of this project to undertake a quality assessment of individual articles; we instead focused on describing the nature of the included studies. 23

| Synthesis of results
The review findings are mapped and narratively synthesized against five distinct food environment research areas: multiple aspects of the food environment, food availability, food composition/labeling, food price, and food marketing. 4Given the large number of studies that measured multiple aspects of the food environment, these studies were further categorized by the method or tool that they employed (i.e., Photovoice, the Healthy Neighborhood Discovery Tool, and Healthy Eating Active Living: Mapping Attribute using Participatory Photographic Surveys [HEAL-MAPPS]) and sociodemographic group studied.

| RESULTS
As Figure 1 shows, the academic database search returned 10,065 potentially relevant results.After 3727 duplicates were removed, 6338 titles and abstracts were screened for relevance, with 460 full texts retrieved and reviewed against the eligibility criteria.Of these, 40 articles were included.The primary reasons for exclusion were that studies did not focus on food environment research 4 or that people were not actively engaged in the project.Two additional articles were included from grey literature, resulting in a total of 42 articles.

| Crowdsourcing data on multiple aspects of food environments
Thirty-four studies used crowdsourcing methods to survey multiple aspects of food environments.Photovoice 64 was the most popular method of doing this (n = 25 studies), [26][27][28][29][30][31][32][33][34][35][36][37]42,[45][46][47][48][49][51][52][53][54][55]58,60 followed by the "Our Voice" Discovery Tool 65 developed by Stanford University (n = 5 studies), 18,19,38,56,61 and the HEALMAPPS 66 (n = 4 studies). [39][40][41]43 Photooice Photovoice is a visual community participatory action method that enables participants to autonomously capture images and reflect on the strengths and concerns within their community.64 Photovoice enables individuals' perspectives and voices to be represented through images, often capturing the voices of people experiencing disadvantage. 67Participants in the 25 included studies were given disposable or digital cameras, or opted to use their own cameras, and were instructed to take images of their food environments over 4-10-week periods.Participants then completed a focus group or interview, sometimes selecting only one to five of their most representative images for discussion. Throughout thestudies, sessions were run to help guide photography, including prompting images for food insecurity or food environment assets and barriers.][27][28][29][30][31][32][33][34][35][36][37]42,[45][46][47][48][49][51][52][53][54][55][56]58,60 Low socioeconomic position.Eleven studies used Photovoice to crowdsource images from people in middle-to low-socioeconomic positions (SEPs). Six of tstudies focused on children from low SEP households, [27][28][29][30][31]42,60 two on adults in low SEPs, 33,53 one on comparing food environments across areas of low and high SEP in one city, 54 and one on comparing food environments between two cities with similar SEPs.55 Using Photovoice, there was a common consensus across most studies that healthy food was not affordable or accessible to participants and their families.[26][27][28][29][30][31][32][33][34][35][36][37]42,[45][46][47][48][49][51][52][53][54][55][56]58,60 Studies capturing children's food environments using Photovoice developed unique ways of keeping children engaged, including supermarket shopping trips, staging images of their food environments at home, and creating a fictional story about their food environments.28,31 In a study by Bosco et al., 27 38 children took images using GPS-enabled cameras provided of the food they came across over a 24-h period and then discussed each image in an interview.Bosco's results emphasized the independence that older children have regarding food consumption.Children additionally highlighted in interviews the influence that social media and advertising have on the types of foods that they consume. Children also indicated in interviews that fast food stores are located close to home 60 and that supermarkets with a healthier selection were located in more affluent areas that were difficult to reach by public transport.29 Leung et al.'s 42 participants highlighted the abundance of unhealthy food and marketing influences on eating behavior, particularly at school and around their neighborhood.Diez et al. and Lindow et al. 33,53 both crowdsourced food environment data from adult populations.Lindow et al.'s 33 participants highlighted the shame and guilt they felt when they were food insecure.In interviews, the parents explained they took their children out to fast food restaurants for special occasions to create a sense of normality for their children.Diez et al. 53 to assess demographic characteristics to be included in the study; and (2) to assess technology ownership, diet apps, or prior nutrition education.Participants then rated 10 foods using the traffic light approach to assessing nutritional quality.
More than 50% of participants accurately categorized all 10 foods.
Green foods received the highest food accuracy scores with a mean score of 99.7%, followed by yellow (68.8%) and red (68%).
Crowdsourcing food sources within food environments The official public database of this community had reported that there were almost 50 grocery stores available to its residence; however, the data from this project revealed that there were only three grocery stores and that most were actually liquor stores.The aim of YPAR 2.0 is to bring attention to areas of decline for the presence of smaller stores and markets provided access to fresh food and vegetables, supermarkets used price promotions to increase consumption of unhealthy options.Gravina et al. 55 compared two cities (a major city and an industrial city) with similar SEP and highlighted the low cost and high availability of fast food.Participants from the major city indicated that price influenced food purchase rather than quality of food compared with the industrial city, which had high quality and affordable options. 54Additionally, in another study, Gravina et al. 55 highlighted the economic disparities experienced by people in different socioeconomic groups in a single city.
Remoteness.Four studies used Photovoice to crowdsource images of the food environment from rural locations, 26,34,35,46 with two focused on children. 26,35Neill et al. 46 found that participants needed to stockpile food with a long shelf-life as the weather sometimes prevented them from accessing food.Mabry et al. 34 explored the importance of the home environment in maintaining a healthy diet.Balvanz et al. 35 crowdsourced images from African American children living in a rural area who reported that the cost of travel and a reliance on others (such as family, school, or relatives) to acquire food were barriers to accessing healthy foods.At the end of the study, a community forum was held to present study findings to community stakeholders and leaders to assist in generating support to implement action steps and give a platform to the barriers participants experience to accessing healthy food. 35In a study by Findholt et al., 26 nine 15-18-year-old children collected images of the characteristics within their community that affected food choice.
A consensus among children reported that there is a lack of fresh fruits and vegetables in rural grocers in the United States, with one participant photographing less than two dozen vegetables on a supermarket shelf. 26rst Nations adults.Three studies crowdsourced food environment images from First Nations participants in Canada (n = 2) 47,49 and Australia (n = 1). 58All three studies emphasized the importance of traditional foods in First Nations culture and diet; however, Hanemaayer et al. 49 highlighted difficulties in acquiring traditional foods because of their high cost.Study participants in both Adams et al. and Lardeau et al. 47,58 faced food insecurity, depicting empty fridges and a reliance on food banks.
Migrant populations.One study used Photovoice to capture the food environments from the perspectives of migrants to Canada from South Korea, Haiti, and parts of West Asia. 51Rodriguez found that participants faced many barriers to accessing healthy foods, including the distance to stores and the high cost of healthy options. 51oader community groups.Four studies engaged volunteers from across entire communities. 32,36,45,52 access to healthy food more broadly in the community and school.
Belon et al. 45  invited policymakers to become part of the experience by including them as study participants. 36Results from the follow-up Photovoice study displayed that residents saw an increase in fresh food availability across all six communities.Additionally, some communities introduced policies to facilitate healthy food environments, including restricting unhealthy food around schools and reducing barriers to increase urban agriculture. 36oader groups of children.Two studies crowdsourced images from children's school food environments. 37,48Both studies emphasized children's desire to be heard and have power in deciding what foods are available in their cafeteria. 37,48Spencer et al. 48reported that there were limited menu options and unappetizing foods in the school cafeteria and that children often sourced food off campus.Harper et al. 37 followed the journey of food from farm to plate.At the conclusion of the study, the project was used to advocate for the ongoing supply of food to schools from local farms.

The Healthy Neighborhood Discovery Tool
The Healthy Neighborhood Discovery Tool is a modified version of Photovoice administered through an electronic tablet to collect various aspects of food environments. 65Six studies used the Discovery Tool in the USA (n = 4), 18,19,38,42 Australia (n = 1), 56 and Colombia (n = 1). 61Studies had 8-97 38,61 participants partaking, resulting in the collection of between 127 and 820 56,61 images and 96-633 audio narratives. 38,619 found that participants traveled outside of their neighborhood to obtain affordable healthy food.Tuckett et al. 56 modified the Discovery Tool to allow older adults to identify enablers and barriers to healthy eating in the food environment while concurrently collecting images.
Sheats et al. 18 also surveyed the food environment from the perspective of older adults, with both studies finding that participants struggled to locate and afford foods for special dietary requirements (e.g., diabetic or high blood pressure diets). 18,56González et al. 61 found issues in school environments in which cafeterias did not offer vegetarian options.
Healthy Eating Active Living: Mapping Attribute using Participatory Photographic Surveys (HEALMAPPS) HEALMAPPS was used in four studies to facilitate community efforts to create environments that support healthy eating. 66This method includes photomapping, group decision-making, and mobilizing community conversation and action.41]43 All four studies indicated that the availability of healthy food was low, [39][40][41]43 particularly in school environments where sugary snacks and beverages are readily available. 41 ee studies were conducted in rural settings.39,41,43 Despite having many farmers around the local community, Lindsey et al. 43 highlighted how their fresh produce was sold to large retailers because of the limited economic feasibility of operating stores in rural areas.This study also found that four communities often shared food resources, for example, by hosting farmers' markets. Jon et al. 39 recruited 241 people across 21 communities to take images of their food environment. This sudy found that the geographical isolation of some rural communities can affect the cost and availability of food, with one community reporting they only had one grocery store in town that often had high prices.39 At the conclusion of each study, community meetings were held to gather key stakeholders and start conversations on the need to address the identified barriers to healthy eating through policies and local actions.

| Crowdsourcing food sources within food environments
Three studies crowdsourced data on the availability of food stores within two communities in the USA 24,44 and one community in Canada. 50Gomez-Lopez et al. 44 compared three different full-line grocery store datasets (grocery stores that sell fresh produce, such as fruit, vegetables, meat, and dairy).These included (1) a community crowdsourcing initiative, (2) a national database, and (3) Yelp (a social media platform and online directory crowdsourcing ratings and reviews of local businesses). 68The community initiative identified that only 24% of the census tract neighborhoods in the area, compared with 60% with the national database, had at least one open full-line grocery store.Fast and Rinner 50 uniquely aimed to map the availability of food stores by creating a volunteer geographical information system (GIS).The project was unable to generate a dataset that was comprehensive enough for the region to conduct a spatial analysis using the 227 image contributions by volunteers; however, it uncovered innovative community food assets such as seed libraries, pollination gardens, and publicly available fruit trees. 50Akom et al. 24 created a digital method for mapping the real-time availability of food using the platform, Streetwyze.The study found that while official public databases reported that there were almost 50 grocery to occur efficiently and in real time.The public availability of project data is a key principle pertaining to high-quality citizen science. 22oniverse is a popular open access platform in other citizen science disciplines and has been a central asset in crowdsourcing agriculture, marine, and planetary research. 13,14,73However, our review only found one study that used the platform to crowdsource data on food environments. 59Other studies opted to use other established open platforms such as Crowdmap and Streetwyze, 24,50 which also enable geo-location tracking of images.
The availability of open access data is a key tenet within the principles of citizen science. 22Of the 42 articles we reviewed, only six 24,50,54,57,59,60  discussions on the healthiness of food environments.This review found evidence (albeit from only four studies) of studies creating big data or using big data sets that had been crowdsourced for other reasons to crowdsource food composition, labeling, and pricing data. 57,62,63However, our scoping review also raises ethical questions about mass data collection via crowdsourcing.For example, during recruitment, researchers should obtain informed consent for ongoing public use of images and information that citizens share. 74Furthermore, researchers need to consider how citizen scientists will be appropriately acknowledged for their contribution towards the outcome of the study, aligning with a key principle of citizen science. 22

| Community empowerment
Another central feature of citizen science is actively involving community members in the scientific endeavor. 22Crowdsourcing studies, particularly Photovoice studies, were focused on empowering citizens to collect scientific data about their own food environments and be advocates for change. 67Importantly, 29 of the included studies crowdsourced data from priority populations, thereby engaging and amplifying the voices and experiences of priority groups that experience social and/or economic exclusion to advocate for healthier food environments.Given that people who experience the greatest disadvantage are least likely to be represented in public policymaking, Photovoice and other active crowdsourcing approaches may represent a feasible and acceptable method for engaging priority populations in decision-making.Photovoice and other active crowdsourcing approaches should continue to be considered in efforts to reduce inequities in diet-related health. 75This aligns with one of the original intentions of participatory research, which is to reduce power imbalances in decision-making-including whose expertise and knowledge gets heard and prioritized.Community-based participatory research approaches such as Photovoice have a large emphasis on empowering people with lived experience to share their stories and experiences with researchers in a way that attempts to mitigate the existing power imbalance between participant and researcher and position community members as the experts.
1][62][63] This points towards the potential to use crowdsourcing to streamline the monitoring of food environments in low-resource settings, which remain relatively underexplored compared with higher-income settings despite having higher burdens of morbidity and mortality from NCDs. 76other principle of citizen science is receiving feedback from the project, including how data is used within research and social and political outcomes as a result of the project. 22Therefore, a key ethical consideration concerning the ongoing engagement of community members, including priority populations, is the need to conduct more follow-up studies that examine the impact of crowdsourcing approaches on policy and community outcomes.Moreover, researchers need to consider how they will maintain communities' connections with their local food environments if the data they are using are not collected locally.Without this evidence, it will be difficult to ascertain whether communities are in fact being "empowered" to become agents of change or tokenistically consulted. 77The latter is problematic as it does not disrupt the status quo but rather enables power imbalances in decision-making. 77Table 2 summarizes the implications of our findings for future research and practice.

| Strengths and limitations
A key strength of this scoping review is our use of a comprehensive search strategy that was applied across six electronic databases and had input from experts in the fields of systematic reviews and citizen science.Citation manager (Endnote) and systematic review (Covidence) software were used to ensure data searching, screening, and extraction were rigorous and transparent.It should be noted that we identified some inconsistencies in the definition of "crowdsourcing" as we were screening the search results for this review.Crowdsourcing is often defined in relation to active (e.g., citizen science) or passive (e.g., secondary data) participation in research.Given that our focus stemmed from a citizen science perspective, whereby the first principle of citizen science is to "actively involve citizens in scientific endeavors," 22 we excluded passive crowdsourcing studies from this review.While some passive approaches may also constitute valid ways of collecting "big" food environment data, it is arguable whether such studies can be defined as crowdsourcing (because of the absence of active citizen participation).Our review, therefore, makes a novel attempt to define crowdsourcing for food environment research.Additionally, while third-party platforms that host and crowdsource user-generated content, such as Yelp, could be considered as tools to crowdsource within the food environment, they are not considered citizen science in our review.This is because these platforms do not inform contributors about the uses of the data provided outside of the platform, such as for research purposes, and therefore could not be included. 22

| CONCLUSION
To date, many studies have piloted small-scale crowdsourcing projects to understand multiple aspects of our food environments and their relevance to health through the eyes of citizens and comunity membersincluding those who experience the most social and/or economic disadvantage.This evidence suggests that crowdsourcing can provide a platform for citizens to have their voices and local knowledge heard.
Fewer projects have involved developing the infrastructure required to actively engage large numbers of citizens and collect and process large amounts of data.Additional research will be needed to capitalize on the growing potential to develop scalable crowdsourcing platforms that are relevant to public health nutrition.Such crowdsourced data may improve public and policy engagement with key food policy actions that are required to create healthier food environments for all.
T A B L E 2 Future applications of crowdsourcing to create healthier food environments.
Citizen engagement • Citizens can be better engaged if they know their contributions are part of something bigger, such as policy development or action.• Researchers should liaise with external community stakeholders (e.g., community organizations and champions) or existing projects to work in partnership and facilitate engagement and action.• Traditional recruitment methods such as flyers, email lists, and social media should be explored and optimized.
Tailor your approach for your population group • Use methods that will enable you to see and understand participants' perspectives of food environments in a way that is easy and accessible to them.• While there is potential for ongoing research and technological advancements in the field of crowdsourcing, approaches that actively capture and amplify the lived experiences and voices of community members should be prioritized.

Training and empowering participants
• Researchers should invest in training community members on how to best execute their role as co-researchers (this should contribute to their skill development and capacity).• If researchers cannot provide training to community members, simple written manuals or in-app instructions are key to guiding participants to effectively collect data for a project.• New technological advancements should continue to be explored and evaluated to facilitate food environment data collection in a way that empowers people to be agents of policy change in their communities.
crowdsourced images from a diverse range of people aged under 24 to over 65.The study highlighted the complex nature of food environments, showing how the cost, availability, and accessibility of foods and beverages influence eating decisions.Diez et al.52 uniquely gathered data and participants from two previous Photovoice studies to identify 12 recommendations to improve the healthiness of food environments in their community, for example, by increasing the availability of healthy foods and restricting vending machines at work sites.Kramer et al.36 used a longitudinal pre-post study design across six communities to evaluate the change in food environments as a result of an initial Photovoice study.Kramer et al.

4. 1 |
Citizen science, crowdsourcing, and big dataThe increased access to mobile phones and the internet has catalyzed the emergence of innovative data collection techniques.This growth and access to technology have alleviated some of the burden of collecting large amounts of data, such as for administering crowdsourcing projects within the community.Citizen participation via these digital tools and platforms enables data collection (and sometimes analysis) acknowledged that while F I G U R E 1 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) chart of article screening and selection process.T A B L E 1 A detailed description of all articles included in this scoping review.
Four themes were identified: (1) food characteristics; (2) social environment; (3) kitchen, cooking, and dining environments; and (4) food insecurity.Participants highlighted the role family plays in food preparation and cooking.Most of these children were food insecure and stated they often took extra food home from the afterschool program to make T A B L E 1 (Continued) This study showed the independence of this age group when it comes to food consumption and purchasing.Students were often also influenced by images of food shared via social media and social influences and discussions.Findings identified a lack of availability of healthy foods at school, with students often opting to venture off campus for fast food, suggesting that cafeteria food was less than satisfactory.Participants took a total of 233 photographs across both cities (163 in the major city and 70 in the industrial port city).While both cities included access (Continues) T A B L E 1 (Continued) Four major themes were identified: (1) the easy access and availability of fast food at a low cost promote the consumption of unhealthy food; (2) the cost of buying food is a struggle, with parents shopping around for price promotions or opting to go to discount stores for groceries; (3) parents felt shame and guilt about their experience with food insecurity, particularly about how difficult it was to provide fruit and vegetables to their children; (4) parents expressed using unhealthy foods as a special treat for their children to try to shield them from food insecurity and create some normalcy.T A B L E 1 (Continued) T A B L E 1 (Continued) T A B L E 1 (Continued) (Continues) T A B L E 1 (Continued) T A B L E 1 (Continued) Fitzgerald et al. 32crowdsourced data from both children and adults, finding that while residents had access to urban gardens in the community, there was still a lack of T A B L E 1 (Continued) 38risinger et al.38identified three priority