The Good Food Junction: a Community-Based Food Store Intervention to Address Nutritional Health Inequities

Background This is a 2-year study to assess the early impacts of a new grocery store intervention in a former food desert. Objective The purpose of the study is to understand the early health effects of the introduction of a large-scale food and nutrition-focused community-based population health intervention, the Good Food Junction (GFJ) Cooperative Store, in a geographically bounded group of socially disadvantaged neighborhoods (the “core neighborhoods”) in a midsized Canadian city. The GFJ grocery store was tasked with improving the access of residents to healthy, affordable food. The 5 research questions are: (1) What is the awareness and perception of the GFJ store among residents of the core neighborhoods? (2) Are there differences in awareness and perception among those who do and do not shop at the GFJ? (3) Will healthy food purchasing at the GFJ by residents of the core neighborhoods change over time, and what purchases are these individuals making at this store? (4) What early impact(s) will the GFJ have on key health-related outcomes (such as household food security status, vegetable and fruit intake, key aspects of self-reported mental health, self-reported health)? and (5) Are the effects of the intervention seen for specific vulnerable population groups, such as Aboriginal people, seniors (65 years old or older) and new immigrants (settled in Saskatoon for less than 5 years)? Methods The research project examined initial impacts of the GFJ on the health of the residents in surrounding neighborhoods through a door-to-door cross-sectional survey of food access and household demographics; an examination of GFJ sales data by location of shoppers' residences; and a 1-year, 3-time-point longitudinal study of self-reported health of GFJ shoppers. Results Analyses are on-going, but preliminary results show that shoppers are using the store for its intended purpose, which is to improve access to healthy food in a former food desert. Conclusions To our knowledge this is the first large-scale study of a full-service grocery store intervention in a former food desert in Canada that has used multiple data sources, as well as longitudinal analyses, to examine its effects. Its findings will contribute significantly to the knowledge base on food environment interventions.

The Good Food Junction: A Community Based Food Intervention to address Nutritional Health Adequacies is an evaluation of an intervention project called the "Good Food Junction", a grocery store/health center developed to provide healthy food options in a Saskatoon food desert.The proposal seeks to evaluate the impact of the store on residents in surrounding neighbourhoods in order to better understand if and how interventions such as the one under study can lead to increased availability and purchasing of healthy food among both residents in the areas and users of the store specifically, and improved health related outcomes, in general and among vulnerable populations.
To do this, the researchers propose a cross-sectional survey of 600 households sampled door to door in the selected neighbours, a longitudinal study of 300 households drawn from the survey followed up at 6,12,and 18 months.The longitudinal survey will include measures from the CCHS, as well as specific questions on grocery shopping.Food purchasing data collected from GFJ will also be tracked over time.The food tracking data will also be used to evaluate a number of in-store interventions funded elsewhere.
A balanced team with experts in epidemiology, survey methods, nutrition, food programs, aboriginal health who have generally had experience working together.There is evidence of strong partnerships with community and provincial decision makers.Similarly the cooperation between the good food junction, the Station 20 West Centre, and other partners is well established and active.The PI team is a mixed of an early career researcher with a more established investigator.Clearly, the investigators are strong supporters of the intervention and programming and may benefit from voices outside of Saskatoon who may be less intimately involved.Otherwise, the environment is a supportive one for this research.
The issue of food deserts in communities is an important one and many communities are interested in addressing the issues of food security and healthy food availability.Systematic assessment and evaluation of community level programming is also more rare than it should be, particularly in the attempt to link programming and intervention to individual health outcomes.This proposal aims to provide information that will assess the impact of the specific program but should help other communities in the implementation of food availability interventions.Furthermore, most work has taken place outside of Canada.The work suggested here is well justified as important and novel.
Background literature supports the need for the study.The review is broadly supportive of an effect; The Good Food Junction: A Community-Based Food Intervention to Reduce Nutritional Health Inequities however, there is some question over the mechanism linking the availability of food to health outcomes.The nature of this mechanism (and the causal direction) is important for the development and implementation of interventions and it may be worth expanding on the potential for alternate explanations for the association between SES and obesity in particular, and provide further confidence that the intervention would have a causal effect.It should be considered that the potential reason for the lack of effect of previous interventions may be due to a lack of causal effect.
Research questions are in general clear, tied to the conceptual model, and well stated.Research question 4 is somewhat problematic.It is not immediately evident what the difference might be between a health impact and an early health impact.It is important to study the health impacts of these interventions, however, as stated in the baseline report, the effect of the intervention on health may be hard to demonstrate for many years and difficult to tie to a single intervention.While assumptions may be made about intermediate variables (such as additional purchasing of fruits and vegetables), answering research question 4 as part of the study's primary objectives may be challenging.
The piggybacking of intervention studies demonstrates the importance of the research and the value added by this study.However, it would be helpful to understand the specifics of the types of interventions that might be implemented and evaluated here.
This proposal has a number of strengths in terms of the methodology used.The existing baseline data assessment already completely prior to the intervention allows for pre-post comparisons, and demonstrates the ability of the research team.The use of administrative (food purchasing data) to assess changes in purchasing patterns appears to be a rich source of information.The longitudinal design would be effective at assessing changes in behaviour, purchasing, and health outcomes.
A timeline for the research would be helpful.The text suggests that the first wave of the survey will be conducted within 6 to 8 month from the opening of GFJ in September 2012, which suggests that the study will be conducted almost immediately upon the start of the study.Time for recruitment of neighbourhood based research assistants, or development and pre-testing of the survey does not appear to be included.Sampling for the cross-sectional survey follows a study previously conducted in Toronto using residents of the neighbourhood to increase participation rates.The unit of sampling, randomly generated from an electoral list, appears to be a "household."Selection of participants is not unfortunately described.Are measures being collected for every member of the household or just the person who answers the door/person with the next birthday or other method for selecting participants within a household?Similarly, there do not appear to be inclusion or exclusion criteria for participants.For instance, are children and adolescents eligible to participate?As the measures for the longitudinal survey, in particular, are focused at the individual level (such as measured height and weight) rather than the household level.
Sampling will not be based on those who shop at GFJ.However, a stratified random sample would appear to have a number of advantages over a simple random sample.The sample size calculation suggest that half of the households would shop (regularly?ever?) at GFJ, but this figure is not justified in the text.Although the baseline survey in the appendix suggests that 55% would considering shopping regularly, there is a significant possibility that this figure would be lower-particularly if the intervention were less successful.A stratified sample would allow for sufficient sample size of exposed group for this study.Research question 4 moreover suggests stratifying the analysis by vulnerable population group, and sample size calculation should allow for this strata to be sufficiently large to provide useful data.
In the "nested" longitudinal design, a subset of the households are chosen for inclusion but the process for selecting households from the cross-sectional survey is not described.Nor whether the selection process (and/or consent to for follow up) takes place during the initial recruitment after the initial analysis of the cross-sectional data.
The analysis of the use of the food purchasing data (fixed and random effects models) to answer research question two suggests that individual level data will be available.However, this appears to assume that postal codes will uniquely identify individuals and the validity of this assumption needs to be justified.It would be useful to assess the food purchasing habits of the longitudinal sample but this is not described.Changes in food purchasing are likely to be confounded with other effects such as prices and season but it isn't clear if these are to be controlled for in the analysis.The KTE plan is vague; however, it appears that the researchers have considerable experience and interest in ensuring that research uptake occurs.The research team and collaborators also suggests an integrated knowledge exchange process that ensure excellent KTE within Saskatoon/Saskatchewan.Other efforts to expand the impact nationally and internationally would also be valuable.

Research Approach
This proposal will investigate the impact of a newly established population health intervention (i.e. a local neighborhood full-service food store called the Good Food Junction) on the health of people in its surrounding neighborhoods.This will be accomplished by obtaining data from three studies (a cross-sectional household survey, a nested longitudinal study and a food-purchasing tracking study) to answer four research questions on : 1) awareness of the GFJ; 2) trends in food purchasing over time; 3) changes in food purchasing following a number of GFJ-initiated advertising promotions; and 4) impact of GFJ on selected health outcomes, in general, and for selected vulnerable groups.
With respect to the cross-sectional study, it does not appear that the selection of the households will, in fact, be random.The eligibility criteria for participating households has also not been specified.The sample size calculation is unclear and it is difficult to know the magnitude of the 'small effects' that is described.It is also difficult to understand how household food purchasing patterns will take into account household size and other socio-demographic characteristics which influence food purchasing.In assembling the study population, there does not seem to be consideration for how vulnerable groups (Aboriginal people, seniors and new immigrants) will be identified.
For the nested longitudinal study, there is no justification for the sample size of 300 (200 who purchase food at GFJ and 100 who don't).Also, how will the definitions of households who do purchase and who don't purchase food at GFJ be determined?For example, what if only one household member purchases at GFJ? From whom will the Food Frequency Module be obtained?Who will contribute the self-reported information on the health outcomes?Because the detail on methods is insufficient it is difficult to understand how the intended response to the research questions will be obtained.
For the food purchasing tracking study, how the data will be compared to the rest of Saskatoon is unclear.How a stratified analysis will be conducted between residents and non-residents is also unclear.
The strengths of the proposal include 1) its relevance and timeliness in terms of the topic and the recent introduction of a population heath intervention into a new environment ; 2) having the research assistants come from the local neighborhoods; 3) collaboration of the GFJ and local health authorities; 4) a strong

Impact of the Research
It is clear that the introduction of a new source of good food into a previously 'desert area' will have an impact on food access and food consumption patterns.The ability of this research to make a direct link between the Good Food Junction and health outcomes is unclear, given that other similarly new interventions are also at play in the immediate surroundings which also affect these same outcomes (e.g.Food kitchens and mental health).
-14 Operating Grant: Population Health Intervention Research to Promote Health and Health Equity/Subvention de fonctionnement : recherche interventionnelle en santé des populations visant à promouvoir la santé et l'équité en santé