Consumption of a diet rich in fruit and vegetables may help prevent a range of diet-related health problems including CVD, cancer and stroke(Reference Joshipura, Ascherio, Manson, Stampfer, Rimm, Speizer, Hennekens, Spiegelman and Willett1–3). Health promotion efforts to improve intakes have traditionally concentrated on individually focused psychological and educational approaches, but these have met with limited success(Reference Pomerleau, Lock, Knai and McKee4). In recent years there has been a growing recognition that environmental, as well as individual, risk factors may influence food choice and thus nutrient intakes(Reference Booth, Sallis and Ritenbaugh5–Reference Cummins and Macintyre7).
Researchers have documented spatial variations in food consumption patterns, with neighbourhood deprivation independently predicting poor diet(Reference Shohaimi, Welch, Bingham, Luben, Day, Wareham and Khaw8, Reference Forsyth, Macintyre and Anderson9). Differences in the characteristics of the food retail environment between deprived and affluent neighbourhoods have been hypothesised to explain these geographical inequalities. In the USA, presence of a supermarket has been associated with an increase in consumption of fresh fruit and vegetables(Reference Rose and Richards10–Reference Zenk, Schulz, Hollis-Neely, Campbell, Holmes, Watkins, Nwankwo and Odoms-Young12) but the spatial patterning of these stores indicates that residents of minority and deprived neighbourhoods have poorer access to them(Reference Moore and Roux13–Reference Morland, Wing, Roux and Poole15). In the UK, New Zealand and Australia the picture is more mixed, with cross-sectional studies reporting positive(Reference Shohaimi, Welch, Bingham, Luben, Day, Wareham and Khaw8) and negative(Reference Pearson, Russell, Campbell and Barker16–Reference Pearce, Hiscock, Blakely and Witten18) associations between neighbourhood characteristics and fruit and vegetable consumption, and minimal differences in the price and availability of food between deprived and affluent neighbourhoods(Reference Cummins and Macintyre19–Reference Pearce, Witten, Hiscock and Blakely21). Intervention studies have also reported contrary findings, with improvements in local food grocery store provision having either having no effect(Reference Cummins, Petticrew, Higgins, Findlay and Sparks22, Reference Cummins, Findlay, Higgins, Petticrew, Sparks and Thomson23) or resulting in a small positive increase in individual fruit and vegetable intake(Reference Wrigley, Warm and Margetts24).
However, many of these studies have simply quantified access to fruit and vegetables in terms of price, geographic distance to, or densities of, grocery stores selling fresh produce. There are few studies, and none in the UK, that have assessed the quality of fresh produce items within grocery stores. The quality of fresh fruit and vegetables on offer within stores may be a micro-environmental mediating variable that influences the purchasing behaviour of consumers while shopping – if an item is bruised, blemished, misshapen, pitted or moulding it may be perceived to be unappetising and poor value for money and thus is less likely to be purchased(Reference Glanz, Sallis, Saelens and Frank25, Reference Jago, Baranowski and Baranowski26). Such purchasing decisions may, in turn, negatively impact on overall diet quality by ensuring that fewer fruits and vegetables are purchased on shopping trips. One study in Detroit has found that mean quality of fruit and vegetables was lower in poor African-American neighbourhoods and higher in supermarkets(Reference Zenk, Schulz, Israel, James, Bao and Wilson27). Positive perceptions of fruit and vegetable quality were positively associated with a higher intake of these items(Reference Zenk, Schulz, Hollis-Neely, Campbell, Holmes, Watkins, Nwankwo and Odoms-Young12). Differences in fruit and vegetable quality may thus exist by store type and neighbourhood deprivation, which in turn may help explain neighbourhood variations in consumption patterns not explained by differences in price and availability. Here we investigate whether the quality of twelve commonly consumed fresh fruit and vegetable items varies by store type, urban–rural setting and neighbourhood deprivation in Scotland.
Ten study sites were purposively selected to represent the range of socio-environmental settings across Scotland on the basis of Scottish Executive’s Urban–Rural Classification Scheme (SEUR) and the 2006 Scottish Index of Multiple Deprivation (SIMD). The final sample of study sites ensured coverage of the four main environmental settings in Scotland: island, rural, small town and urban. Sentinel sites were initially selected by stratifying all available data zones by the SEUR. Data zones are the core small-area statistical geography used in Scotland. There are currently 6505 data zones in Scotland with a mean population of 778 (range 500–1000).
Each of the 6505 data zones was grouped into three settings: (i) urban (SEUR 1 and 2); (ii) small town (SEUR 3 and 4); or (iii) rural (SEUR 5 and 6). Data zones within each of these settings were then divided into deciles of deprivation using the SIMD, an area-based measure of relative deprivation(28). The SIMD is a publicly available continuous measure of compound social and material deprivation. Within the top and bottom deciles of each of the three settings, one data zone was randomly selected as the nucleus of the sentinel site. For each selected nucleus, additional data zones were added to build an overall sentinel site consisting of contiguous data zones that corresponded to the recognised local community. Six sentinels were initially constructed: (i) urban affluent (Broughty Ferry, Dundee); (ii) urban deprived (Scotstoun/Drumchapel, Glasgow City); (iii) small town affluent (Ellon, Aberdeenshire); (iv) small town deprived (Kilbirnie, North Ayrshire); (v) rural affluent (Haddington, East Lothian); and (vi) rural deprived (Dornoch, The Highlands). However, this process did not select island communities (SEUR rural) and, as expected, the numbers of grocery stores available in some rural and small town settings were too small to conduct meaningful analyses. We therefore purposively selected four further sentinel sites to enhance coverage of the range of settings and boost small numbers of observations. Additional sites selected, on the basis of SEUR classifications, were Eilean Siar & Orkney (islands), Cupar, Fife (small town affluent) and Inverness (urban mixed). In total 205 data zones were selected.
A comprehensive list of the street address and postcode of grocery stores selling food for home consumption (excluding takeaway/fast-food and coffee shop outlets) in the study sites was compiled. Data were initially obtained from industry (Institute of Grocery Distribution) and commercial (Marketscan and Catalist) sources. These data were later supplemented using company websites of the major multiple retailers (Tesco, Somerfield, Asda, Sainsbury and Morrisons), discounters (Aldi, Lidl) and freezer centres (Iceland, Farmfoods), online retail directories (Yell.com) and websites of symbol groups (Spar, Londis, Budgens, Costcutter). In addition, data from local authority registers (The Public Register of Food Premises) were also obtained. Data were combined, de-duplicated and cleaned on the basis of matching address and postcodes. Postcode validity was ascertained by joining the retail data with Ordnance Survey Code-Point information and identifying which postcodes could not be grid-referenced.
In total, 466 unique retail facilities were identified including both permanent and mobile/non-permanent locations such as farmers’ market stalls. Of these locations, twenty-two had a missing, incorrect or incomplete postcode. Postcode errors were resolved by using the Royal Mail’s online address/postcode checker(29) and electronic searches of company websites and directories for thirteen of the twenty-two uncertain locations. The final dataset of geo-coded retail food sources for analysis included 98·1 % (n 457) of the initially identified food retail facilities.
Data on quality of fresh fruit and vegetable within stores
Information on the quality of twelve commonly consumed fresh fruit and vegetables was obtained from in-store visits by trained surveyors to all identified outlets in the food retail census. Data on the quality of produce items were collected as part of a broader project on healthy food price and availability. The quality indicator was included in the Healthy Eating Indicator Shopping Basket (HEISB) tool and consisted of a surveyor-reported visual assessment based on three-point Likert scale: 1 = poor, 2 = medium and 3 = good, using the criteria outlined in Table 1. These criteria mirror the evaluations that consumers typically make when choosing to purchase fresh produce. Items included in the present study were apples, bananas, white grapes, oranges, potatoes, onions, carrots, broccoli, cucumber, round lettuce, red peppers and tomatoes. Data were collected in two phases: October/November 2005 to February/March 2006. Details on the rationale for the items included in HEISB have been reported elsewhere(Reference Anderson, Dewar, Marshall, Cummins, Taylor, Dawson and Sparks30).
Scores assigned: low/poor = 1; medium/acceptable = 2; high/good = 3.
Mean quality scores for individual fruit and vegetable items were calculated by shop type (small food superstore, <3000 sq ft; medium food superstore, 3000–15 000 sq ft; large food superstore, >15 000 sqft; specialist food store with a single function, e.g. greengrocer or butcher; and non-food store, e.g. where food is secondary such as a store selling alcohol), by SEUR category (island, rural, small town and urban) and by quintile of area deprivation. For area deprivation, each data zone in the study sites was assigned a score obtained from the 2006 SIMD(28). Data zones were then ranked and categorised into quintiles (1 = least deprived, 5 = most deprived). Differences between categories were assessed by ANOVA. Accepted level of significance was P < 0·05.
A total of 288 (63·0 %) stores stocked at least one fresh fruit or vegetable item. These stores were located in 177 of the 206 eligible data zones. Table 2 describes the distribution of these stores. All large and medium-sized stores stocked at least one fresh fruit or vegetable item, with 71·8 % of small stores selling fresh produce. The proportion of specialist and primarily non-food stores that stocked at least one fresh produce item was 36·8 % and 38·5 %, respectively. The proportion of stores selling these items by SEUR category was highest in island areas and lowest in urban areas. By deprivation the pattern was non-linear, with the proportion of stores selling fresh fruit and vegetables lowest in the most deprived areas (quintile 5) but highest in quintile 4.
SEUR, Scottish Executive’s Urban–Rural Classification Scheme; SIMD, 2006 Scottish Index of Multiple Deprivation.
Table 3 shows mean produce quality scores by store type. Quality was highest for apples (P < 0·000), potatoes (P = 0·001), onions (P < 0·000) in medium stores; bananas in medium and specialist stores (P = 0·002); carrots in medium and large stores (P = 0·039); and red peppers in large stores (P = 0·018). The same general pattern was observed for the remaining items with medium stores performing the best, although large stores had the highest quality scores for round lettuce, tomatoes and cucumber. In all cases, stores where food is secondary had the lowest quality scores or did not stock particular items.
*Large, >15 000 sq ft; medium, 3000–15 000 sq ft; small, <3000 sq ft; specialist, food store with a single function (e.g. greengrocer or butcher); non-food store, store where food is secondary (e.g. one selling alcohol).
Mean produce quality scores by SEUR category are shown in Table 4. Quality scores were generally good for all items in all categories (>2·15). Four fruit and vegetable items had significant differences in quality by SEUR category: apples, potatoes, onions (P < 0·000) and round lettuce (P = 0·10), with rural or small town areas tending to have the highest-quality produce. For items with non-significant differences, rural and small town settings performed the best with the exception of broccoli (island). For nine of the twelve items the lowest mean quality scores were found in urban settings, with the exception of white grapes (rural), round lettuce and red peppers (island).
SEUR, Scottish Executive’s Urban–Rural Classification Scheme.
Table 5 shows mean produce quality scores by SIMD quintile of neighbourhood deprivation, with mean quality scores generally high in all quintiles (>2·08). Of all the twelve items, eight had the highest mean quality scores in SIMD quintile 2 and nine had the lowest mean quality scores in SIMD 4 or 5. Significant differences were found for apples (P = 0·016), potatoes (P = 0·007) and onions (P = 0·001), with highest mean quality scores found in SIMD quintile 2 and the lowest in SIMD quintile 5.
SIMD, 2006 Scottish Index of Multiple Deprivation.
*1 = least deprived; 5 = most deprived.
Previous research undertaken in the UK has suggested that there are minimal differences in food price and availability between richer and poor neighbourhoods(Reference Pearson, Russell, Campbell and Barker16, Reference Cummins and Macintyre19, Reference Cummins and Macintyre31) and as such these environmental factors may not explain geographical inequalities in diet and related diseases. However, the data presented here suggest that the quality of fresh fruit and vegetables within food stores differs by store type, urban–rural context and by neighbourhood deprivation. Overall the quality of fruit and vegetables within the surveyed stores was high. In general, medium-sized stores, stores in small town and rural areas, and stores in more affluent areas tended to have the highest-quality fresh fruit and vegetables on offer. Conversely, stores where food is secondary (non-food), stores in urban settings and stores in less well-off areas tended have the lowest-quality fresh fruit and vegetables on offer. Although not all differences in quality were statistically significant, the patterns were consistent for the majority of the twelve fruit and vegetable items.
These data suggest a possible role for relative quality of fruit and vegetables as a micro-environmental variable which may mediate individual purchasing behaviour. It may be that although the aggregate cost and availability of individual food items does not vary by neighbourhood the quality of items sold does, with quality influenced by the type of store from which food is sold, whether the store is located in an urban or rural neighbourhood and how deprived that neighbourhood is. The patterns observed here build on earlier findings that fruit and vegetable quality is patterned by neighbourhood deprivation and by shop type(Reference Zenk, Schulz, Hollis-Neely, Campbell, Holmes, Watkins, Nwankwo and Odoms-Young12, Reference Zenk, Schulz, Israel, James, Bao and Wilson27). For example, the findings reported here mirror those from an earlier study in Chicago(Reference Block and Kouba32) which found that ‘independent’ and ‘liquor’ stores in an urban area tended to stock the poorest-quality fresh produce. Analogous shops in the present study are ‘small’ and ‘non-food’ (where food is secondary) stores.
Factors influencing quality in Scotland might include: the quality of items purchased from wholesale markets; travel time from wholesale to retail premises (ripening and deterioration); food storage conditions in-store and in-transit; and volume turnover of food items. These in turn may depend on deprivation, rural–urban location and store type. To lessen variations in quality, Scottish retailers and wholesalers may need to improve storage conditions allied with better marketing to encourage faster turnover of perishable goods. Finally, consumers, particularly those on low incomes, seek value for money. If increases in consumption are to be achieved, efforts to improve quality may in turn enhance perceptions of value and retail store reputation, providing a greater incentive to purchase and consume fresh fruit and vegetables from local outlets.
The study outlined here has certain limitations. The three-point Likert scale for assessing fruit and vegetable quality may not have been sufficiently sensitive to capture the full range of variation in item quality, resulting in higher than expected mean quality scores. The scale was a subjective rather than an objective measure although clear guidance on how to rate fruit and vegetable items was given. We were unable to examine inter-rater reliability as only one observer was sent into each store. The study was cross-sectional and ecological in design and does not link fruit and vegetable quality to individual purchasing or consumption behaviours. A snap-shot study such as this assumes that quality is stable over time when in fact it may vary due to other external factors such as wholesale supply. In this context, the study is best viewed as hypothesis generating rather than confirmatory. Further studies that explicitly investigate the influence of food quality in grocery stores on individual purchasing and consumption patterns are required.
The role of food quality, as opposed to price and availability, has yet to be fully investigated in the study of environmental determinants of diet. Although the present study only allows speculation on the possible influence of the patterns reported here, these data suggest that the quality of fruit and vegetable items may be worth investigating as a plausible micro-environmental determinant of purchasing and consumption behaviour. The study provides evidence that poorer-quality fruit and vegetable items are found in stores in urban settings, in stores where food is secondary and in stores in more deprived neighbourhoods. These variations in food quality may help partially explain neighbourhood differences in food consumption patterns.
The present work was supported by the Food Standards Agency (Scotland) as part of the project ‘Accessing Healthy Food: A National Assessment and Sentinel Mapping Study of Food Retailing in Scotland’ (Ref: S04005). D.M.S. is supported by the award of the Philip Leverhulme Prize to S.C. S.C. is also supported by a NIHR Fellowship. There are no conflicts of interest. S.C. wrote the paper and directed analyses. D.M.S. conducted analyses and contributed to the writing. J.D., D.M., A.S.A. and L.S. contributed to paper drafting. M.T. directed and undertook the field work.