Current recommendations in the USA call for increasing the proportion of fruits and vegetables in the daily diet as a way to meet nutritional needs while maintaining energy balance(1). Fruits and vegetables have a high water content and low energy density which can lead to feelings of satiety that reduce energy intake, a key factor in preventing weight gain(Reference Drewnowski2). Additionally, diets high in fruits and vegetables are lower in saturated fat and total fat, and have been associated with reduced risks of CVD and many site-specific cancers(Reference Van Duyn and Pivonka3).
Achieving these goals may be easier for some than for others. Socio-economic status (SES) plays a major role in determining a person's ability to obtain a healthy and health-promoting diet. People with lower incomes tend to report diets that include fewer fruits and vegetables(4). The cost of fruits and vegetables relative to other foods is one factor that can help explain lower rates of consumption by low-income individuals(4–Reference Drewnowski and Darmon7). There may be other factors, however, that are also important.
Recent attempts at understanding the complex association between individuals and dietary composition have focused on the importance of the built environment(Reference Booth, Pinkston and Poston8–Reference Lopez and Hynes14). Neighbourhood characteristics are, themselves, a function of socio-economics and play a key role determining the availability of fruits and vegetables. Neighbourhood SES has been associated with the presence and number of supermarkets where residents are provided a broad range of fruit and vegetable options at lower costs than typically found at small groceries and specialty stores(Reference Morland, Wing and Roux13, Reference Latham and Moffat15–Reference Powell, Slater and Mirtcheva17).
While previous studies demonstrate the advantage of living in higher-SES neighbourhoods, they do not necessarily explain the association between individuals, their environments and dietary patterns. Low-income shoppers are quite adroit when it comes to maximizing their food budgets and feeding their families. They engage a range of practices that include travelling to nearby, more affluent areas and shopping in more than one store(Reference Hersey, Anliker and Miller18–Reference Inagami, Cohen and Finch20). In fact, most low-income shoppers do not perceive themselves as ‘cut off’ from supermarkets. In recent surveys, the majority of low-income participants reported easy access to a supermarket – including many who had no access to a car(Reference Rose and Richards21, Reference Dibsdall, Lambert and Bobbin22).
Less is known about how characteristics of the neighbourhood food shopping environment influence fruit and vegetable purchases and dietary intake. In addition to costs and availability, low-income shoppers are also concerned with convenience, quality and selection. Convenience to these shoppers is not limited to geographic proximity, but also includes notions of temporal proximity and time savings (e.g. shopping at stores that offer products other than foods or planning routes that allow for multiple stops)(Reference Webber, Sobal and Dollahite19). Low-income shoppers also seek variety and quality; they associate quality with freshness, and they express concerns about the quality of produce in smaller stores(Reference Webber, Sobal and Dollahite19). A study of primarily low-income African-American women in a predominantly urban setting found that selection and quality of fresh fruits and vegetables directly affected dietary intake and mediated the relationship between store characteristics (store type, size) and dietary intake(Reference Zenk, Schulz and Hollis-Neely23). Factors like quality and selection are more likely to vary from store to store than across neighbourhoods. Research conducted in the UK found only modest variation in the quality of fresh fruits and vegetables across a range of socio-economic neighbourhoods, with substantially more variation in quality among store types(Reference Cummins, Smith and Taylor24).
The objective of the present study was to examine the relationship between residents’ perceptions of their food shopping environment and dietary intake of fruits and vegetables. Data for these analyses come from the baseline assessment of a social marketing campaign conducted in six inner-city Chicago neighbourhoods, the 5-4-3-2-1 Go! Campaign(Reference Evans, Necheles and Longjohn25). The baseline survey included a set of questions related to household food shopping practices. We assessed satisfaction with three characteristics of the food shopping environment – convenience, quality and selection – and hypothesized that individuals who reported higher levels of satisfaction with the place where they typically purchased fresh fruits and vegetables would similarly report greater consumptions of fruits and vegetables. We further hypothesized that this association was independent of individuals’ perceptions of the costs of fresh fruits and vegetables.
The 5-4-3-2-1 Go! Campaign was designed to alter parental behaviours within ‘obesigenic’ environments in low-income Chicago communities(Reference Evans, Necheles and Longjohn25). As described in detail elsewhere, the campaign delivered an integrated set of science-based messages about nutrition and physical activity through use of small-scale, local media and grassroots efforts such as neighbourhood newspapers, radio advertisement and health fairs. Messages promoted the benefits of a healthy diet rich in fruits and vegetables and the use of community resources that support an active lifestyle(Reference Evans, Necheles and Longjohn25).
The analyses presented here are based on data from 495 respondents who participated in baseline data collection for the 5-4-3-2-1 Go! Campaign and provided complete information on all analytic variables. Based on Census tract data, we identified all households in the six study communities to construct the sampling frame. To increase the efficiency of data collection, we obtained commercially available data on household purchase decisions to develop a more targeted list of households likely to include young children. We randomly sampled from the targeted and general lists to obtain our sample. We proportionately sampled from each list as the recruitment progressed and ultimately exhausted all listed households. We used a standard recruitment script that described the overall purposes of the study and its intended use prior to admitting consenting participants into the study. To verify that parents were eligible for the study, we made initial screening calls to verify that they had children in the 3- to 7-year-old target age range for the 5-4-3-2-1 Go! Campaign and recruited those parents who met our criteria.
Respondents completed a ninety-six-item, interviewer-administered questionnaire approved as part of the overall study plan by institutional review boards at the George Washington University and Research Triangle Institute. The instrument was pretested with eighteen respondents from the targeted neighbourhoods, who were debriefed about item comprehension and not included in the subsequent evaluation. The final instrument incorporated revisions based on pretesting.
Baseline data collection was done in the winter of 2007–2008. The instrument contained a battery of questions on demographics; parent physical activity and exercise knowledge, attitudes and behaviours (KAB); parent nutrition KAB; (parent reports of) child nutrition, physical activity and exercise KAB; parent and child media habits; and social environmental (community resources, safety, crime) variables.
Demographic variables included race/ethnicity (white, black or Hispanic), gender, age, and participation in food assistance programmes (Special Supplemental Nutrition Program for Women, Infants, and Children or Supplemental Nutrition Assistance Program) as a proxy for low SES.
Fruits and vegetables index
Daily intake of fruits and vegetables was derived from a set of four items from the US Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System (BRFSS). The BRFSS items ask respondents to provide frequency counts of: (i) green salads; (ii) servings of vegetables (excluding salad); (iii) servings of fruit; and (iv) servings of fruit juice. Respondent's answers were rendered into counts per day and summed to provide a total estimate for daily intake. In the present analysis, we dichotomized respondents into those who ate three or more fruits and vegetables daily and those who ate fewer than three fruits and vegetables daily based on the median value of the index.
Fresh fruit and vegetable shopping environment
Respondents indicated whether they primarily acquired fresh fruits and vegetables at (i) a supermarket, (ii) a farmers’ market or (iii) a local co-op. For this analysis, we created a dichotomous variable that indicated either supermarket or farmers’ market/co-op. In the USA, local co-ops (cooperatives) tend to be small and are retail outlets that are owned or operated by neighbourhood residents for the mutual benefit of the resident/owners. In comparison, US supermarkets are large, for-profit, commercial enterprises that sell a wide variety of foods and household goods. However, no definitions were provided to respondents as part of the data collection protocol. For fifty-two of the respondents who noted an ‘other’ source of fresh fruits and vegetables, the free-form response was compared with available information on retail food shops and included in one of the above three groupings.
Three items assessed respondents’ perceptions of the place where they primarily acquired fresh fruits and vegetables. The first item asked respondents if they felt the place where they shop for fresh fruits and vegetables was convenient to their home. The second item asked respondents whether they were satisfied with the quality of fresh fruits and vegetables where they primarily shop. The third item asked if they were generally satisfied with the selection of fresh fruits and vegetables where they shop. For each item, a 4-point response set ranged from ‘strongly agree’ to ‘strongly disagree’; responses ‘no opinion’ and ‘don't know’ were converted to missing and excluded from the analysis.
Cost of fruits and vegetables
One item asked respondents to what extent they agreed with the statement, ‘Fruits and vegetables are too expensive’. A 4-point response set ranged from ‘strongly agree’ to ‘strongly disagree’; responses ‘no opinion’ and ‘don't know’ were converted to missing and excluded from the analysis.
Food shopping habits
Two variables examined characteristics related to shopping. First, respondents identified the household's primary food shopper as themselves, their spouse or some other adult. For analyses, these responses were dichotomized to ‘self’ and ‘other’. Second, respondents indicated the number times per month they shopped for food for the household.
Prior to conducting regression analyses, a confirmatory principal factor analysis examined the set of survey items assessing satisfaction with the food shopping environment. A factor loading criterion of 0·50 was used to ascertain whether a perceived satisfaction scale could be created from a series of questions related to convenience, quality and selection when shopping for fruits and vegetables(Reference Comrey and Lee26). All three variables met the factor loading criterion and the Cronbach α coefficient of 0·78 met minimum threshold standards (>0·70) confirming a high level of inter-item consistency. The three items in the scale were summed and averaged to retain the range of the original items.
Multivariable logistic regression was conducted by performing the regression of the dichotomized fruit and vegetable index v. the perceived satisfaction index, variables related to the respondents’ shopping practices and the food shopping environment. The regression model also controlled for sociodemographic characteristics. Analyses were conducted using the STATA statistical software package version 11 (2009; StatCorp LP, College Station, TX, USA).
Respondents’ descriptive statistics are shown in Table 1. The sample included 495 of the 526 baseline respondents (94 %); this group provided complete data on all analytic variables. The sample was primarily female (78·8 %), with a mean age of 35·7 (range 18–75) years. Most respondents were African American (44·5 %) or Hispanic (43·0 %). Based on participation in federal food assistance programmes, 48·3 % of the population could be characterized as low income. Table 1 also provides information on the sample's fresh fruit and vegetable intake and shopping habits. Self-reported intake of fruits and vegetables ranged from 0 to 18, with a mean of 5·1 fruits and vegetables daily. In total, 85·5 % of our sample indicated they ate three or more fruits and vegetables daily. Overall, respondents indicated that they participated in food shopping 4·9 times per month.
WIC, Special Supplemental Nutrition Program for Women, Infants, and Children; SNAP, Supplemental Nutrition Assistance Program.
*Total sample included 526 participants; minor variations due to missing items.
†WIC and SNAP are US federal food assistance programmes for low-income individuals and families.
As shown in Table 2, variables assessing the food shopping environment were significantly related to dietary intake of fruits and vegetables. Respondents who reported higher levels of satisfaction were more likely to eat three or more servings of fruits and vegetables daily. Compared with respondents who indicated they were not happy with their fresh fruit and vegetable shopping options, those who agreed that they had convenient access to quality and selection were 2·13 times as likely to eat three or more servings while those who strongly agreed were 4·42 times as likely to eat three or more servings of fruits and vegetables daily. Additionally, those shopping at a local co-op or a farmers’ market rather than a supermarket were 2·77 times more likely to report eating three or more servings of fruits and vegetables daily.
ref., reference category; WIC, Special Supplemental Nutrition Program for Women, Infants, and Children; SNAP, Supplemental Nutrition Assistance Program.
*Quartiles for age: 18–24 years (ref.), 25–34 years, 35–44 years and ≥45 years.
†WIC and SNAP are US federal food assistance programmes for low-income individuals and families.
Our data also indicated that respondents making frequent shopping trips were more likely to report higher fruit and vegetable consumption. Compared with those who made four shopping trips per month (i.e. weekly), those who made six shopping trips per month were 2·10 times as likely to report higher fruit and vegetable consumption. Our findings did not, however, support the hypothesized independent association between perceived cost and dietary intake.
Despite being a primarily minority and low-income inner-city sample, the majority of respondents reported reasonably high levels of fruit and vegetable consumption compared with national averages(4). Additionally, our sample seemed to have very good access to supermarkets and farmers’ markets. This observation concurs with Rose and Richards’ finding that few low-income individuals lack access to supermarkets(Reference Rose and Richards21).
Our data support the hypothesis that perceived satisfaction with the food shopping environment is associated with fruit and vegetable consumption. Among this urban, primarily minority sample, those who felt that the places they purchased fresh fruits and vegetables were convenient and provided higher levels of quality and selection were more likely to eat three or more servings of fruit and vegetables daily. Compared with respondents who were not satisfied, those who agreed that the places they purchased fruits and vegetables provided quality, selection and convenience were more than two times as likely to eat three or more servings of fruits and vegetables daily, while those who strongly agreed with these statements were more than four times more likely to eat three or more servings daily. These findings confirm those reported by Zenk and colleagues(Reference Zenk, Schulz and Hollis-Neely23). Their study, also conducted among an inner-city low-income population, found that perceived quality and selection influenced dietary intake of fruits and vegetables independent of perceived cost.
Additionally, our data indicate that respondents who shopped at farmers’ markets and/or co-ops were 2·77 times more likely to eat three or more servings of fruits and vegetables daily than were respondents who shopped at supermarkets. These associations are independent of sociodemographic characteristics and whether or not respondents viewed fruits and vegetables as too expensive. In fact, perceived cost was not associated with dietary intake among this predominantly minority and low-income audience. Respondents who agreed that cost was a barrier to eating fruits and vegetables did not report lower dietary intake than respondents who disagreed that cost was a barrier.
The present study has several limitations that must be considered when interpreting these findings. First, our data are cross-sectional and observational. We cannot claim, for example, that greater perceived satisfaction with the food shopping environment would lead to greater increases in fruit and vegetable consumption. It is equally probable that individuals with a preference for diets rich in fruits and vegetables seek out stores with better quality and selection. Similarly, the association between shopping at farmers’ markets/local co-ops and dietary intake may reflect the personal preferences of individuals who seek out these specialty establishments because they perceive them as providing greater value.
Second, our study is a secondary analysis and we are limited by the items available in the survey. For example, we do not have information on sociodemographic variables such as educational attainment or household income. These variables have been shown to be associated with fruit and vegetable consumption in previous research(Reference Ard, Fitzpatrick and Desmond6, Reference Zenk, Schulz and Hollis-Neely23, Reference Ball, Crawford and Mishra27, Reference Shohaimi, Welch and Bingham28). The present survey does not allow us to exclude respondents who do not participate in food shopping. Instead, the survey asks who does most of the household food shopping with response options including self, spouse or other adult. Although the primary shopper may have more experience with the food shopping environment, it is quite common today to find that adults share household food shopping responsibilities and the experiences of occasional food shoppers should not be discounted. Accordingly, we chose to control for, rather than exclude, respondents based on this variable. The survey also did not include questions assessing the availability of a car or other forms of transportation. Individuals with more access to transportation would similarly have greater choice of food shopping venues and may have opted to travel further to seek stores that offer selection and quality. It is easy to conceive that individuals with a car would have a different perception of ‘convenience’ than those who are without.
Other variables may lack sufficient differentiation. Our variable assessing the type of food store was limited to three options – supermarket, co-op and farmers’ market – which may not be sufficient important differences in the shopping experience in an environment that includes neighbourhood grocers and corner stores. Future studies should consider the complexity of the food shopping environment. Providing respondents with more options – including corner stores and local groceries – and clearly defining retail outlets to help respondents distinguish among these venues may promote a better understanding of how perceived quality and store type influence affect fruit and vegetable intake.
Finally, it is important to note that the variables measuring dietary intake of fruits and vegetables assess all fruit and vegetable consumption including fresh, frozen, canned and dried forms, while our measures of the food shopping environment ask specifically about the place where respondents typically purchase fresh fruits and vegetables. The place where respondents in our study purchase fresh fruits and vegetables may not be the place they purchase all forms of fruits and vegetables. In fact, previous studies have found that low-income shoppers frequently shop at more than one venue(Reference Hersey, Anliker and Miller18).
Despite these limitations, our study extends the growing body of literature that highlights the importance of personal, subjective assessments of the food shopping environment(Reference Webber, Sobal and Dollahite19, Reference Zenk, Schulz and Hollis-Neely23, Reference Cummins, Smith and Taylor24, Reference Freedman and Bell29). Our analysis examines the role that perceptions of the quality, selection and convenience may play in promoting increased consumption of fruits and vegetables. Our findings suggest that these perceptions matter. Low-income shoppers do not simply make dietary choices based on cost and availability; they also consider less tangible aspects of the food shopping environment that relate to quality and satisfaction.
There is growing support in the public health community for policy, system and environmental changes that promote healthy eating(Reference Sacks, Swinburn and Lawrence30–Reference Glanz and Yaroch32). The goal of policy, system and environmental change is to alter the built environment by decreasing costs and barriers associated with healthy choices. Those pursing policy, system and environmental change have attempted to increase fresh fruit and vegetable offerings in corners stores(Reference Gregson, Foerster and Orr31) and to increase the number of and access to farmers’ markets(Reference Wasserman, Tropp and Lakins33, Reference Briggs, Fisher and Lott34). Providing access is an important step. However, interventionists would do well to think past simple structural obstacles and consider the relational contexts of persons and environments(Reference Cummins, Curtis and Diez-Roux35). Our study offers a cautionary note to those who promote policy, system and environmental change. It suggests that lowering barriers without due attention to the concerns of the target populations is likely to be met with resistance; alternatively, working with local retailers to improve shopping convenience and provide a good selection of quality food items may influence consumers’ overall fruit and vegetable intake independent of price and availability.
The research was funded by a grant from the PepsiCo Foundation and Charitable Contributions. The authors have no conflicts of interest. W.D.E. and J.L.B. contributed to the development of the 5-4-3-2-1 Go! Campaign. All authors shared equally in the conceptualization of the analyses reported herein. J.L.B. and J.S. implemented the analyses. All authors reviewed the analyses and contributed to the development and editing of the manuscript.