Understanding the Intersection of Race/Ethnicity, Socioeconomic Status, and Geographic Location: A Scoping Review of U.S. Consumer Food Purchasing

Disparities in diet quality persist in the U.S. Examining consumer food purchasing can provide unique insight into the nutritional inequities documented by race/ethnicity, socioeconomic status (SES), and geographic location (i.e., urban vs. rural). There remains limited understanding of how these three factors intersect to influence consumer food purchasing. This study aimed to summarize peer-reviewed scientific studies that provided an intersectional perspective on U.S. consumer food purchasing. Thirty-four studies were examined that presented objectively measured data on purchasing outcomes of interest (e.g., fruits, vegetables, salty snacks, sugar-sweetened beverages, Healthy Eating Index, etc.). All studies were of acceptable or high quality. Only six studies (17.6%) assessed consumer food purchases at the intersection of race/ethnicity, SES, or geographic location. Other studies evaluated racial/ethnic or SES differences in food purchasing or described the food and/or beverage purchases of a targeted population (example: low-income non-Hispanic Black households). No study assessed geographic differences in food or beverage purchases or examined purchases at the intersection of all three factors. Overall, this scoping review highlights the scarcity of literature on the role of intersectionality in consumer food and beverage purchasing and provides recommendations for future studies to grow this important area of research.


SUPPLEMENTARY MATERIAL
Three studies examined purchasing using HEI and significant, though conflicting, results were identified [27,53,54].  identified significantly better HEI-2010 scores among Hispanics than NHW [27]. Vadiveloo et al. 2019 also identified better HEI-2015 scores among Hispanic compared to NHW, but only in the southern region of the U.S. [53]. Among regions, they found either significantly worse scores or no significant differences. Vadiveloo et al. 2020 found that compared to NHW, NHB had significantly lower HEI-2015 scores, and NHO had higher scores, while there were no differences between NHW and Hispanic groups [54]. Five studies examined kilocalories purchased and all identified significant cross-sectional or longitudinal differences across racial/ethnic groups [29,42,[47][48][49]. Despite these differences, the associations were inconsistent across studies. Some identified higher energy density purchasing among NHB compared to NHW [47][48][49]; others identified the opposite when examining overall kilocalories purchased) [42]. Among the nutrient purchasing outcomes, four studies examined sugar [43,[47][48][49], three examined saturated fat [47][48][49], and two studied sodium content [48,49]. There was a consistent pattern that NHB had significantly higher purchasing of sugar than NHW across studies; whereas, findings on differences between NHW and other racial/ethnic groups were inconsistent. Two studies examining saturated fat identified significantly lower purchasing among Hispanics compared to NHW [47,49], though differences for other groups were less clear. Stern et al. (2016) and Taillie et al. (2016) examined sodium density of food purchases and identified significantly higher purchasing among NHW compared to NHW across different food purchasing patterns and full-service food retail chains, respectively [48,49].
Four studies examined purchasing outcomes that were not part of our primary outcomes of interest, including total grocery dollars spent and food and beverage purchases with price promotions (e.g., coupons), low-content nutrient claims (e.g., reduced fat, no added sugar), degree of processing (e.g., highly-processed, minimally-processed), and degree of ready-to-eat (e.g., requires cooking, ready-to-heat) [29,47,48,51]. All identified significant differences across racial/ethnic groups. Cullen et al. (2007) identified significantly greater grocery dollar spending on food and beverages among NHW than NHB or Hispanic. Across the other three studies [47,50,51], NHB, Hispanic, and non-Hispanic other (henceforth NHO) demonstrated less purchasing of food and beverage products with price promotions and low-content nutrient claims than NHW. In addition, Asians compared to NHW demonstrated less beverage purchasing with low-content nutrient claims, but more food purchases with low-content nutrient claims and food and beverage purchases with price promotions. Degree of processing and ready-to-eat purchasing also differed, with NHB and Hispanics demonstrating less purchasing of highly processed and ready-to-eat food products compared to NHW; however, NHB also purchased more highly processed beverage products.

Socioeconomic Status
Key findings from studies that evaluated socioeconomic differences in consumer food and/or beverage purchases are summarized here and presented by study in Table S2. Nine studies examined fruit and/or vegetable purchasing [26,29,31,32,34,36,39,45,49] with four identified significant differences [31,34,39,45]. Greater purchases of fruit only, or fruit and vegetables were identified among higher SES groups as compared to lower SES groups. Four studies found no significant differences in vegetable purchasing only, or fruit and vegetables by SES groups [26,29,32,34]. One study examined purchasing of whole grains, and did not find significant differences among SES groups [39].
Seven studies that found significant results, but many inconsistent findings were noted. For example, Grummon et al. (2017) found a statistically significant relationship between SES and salty snacks purchasing, but French et al. (2014) did not [32,34]. Nine studies examined SSB purchasing [26, 29, 30, 32-34, 36, 39, 52] and five examined non-sweetened beverage purchasing [29,31,34,36,52]. Of the studies that assessed SSB purchases, five reported significant differences among SES groups with most studies showing that higher SES groups had lower SSB purchasing or greater reductions in SSB purchasing over time. Among the studies that examined unsweetened beverage purchases, four studies reported significant findings, and three studies indicated that lower SES households purchase less unsweetened beverages [31,34,45], the fourth study was unable to identify the specific differences between the groups [29].
Three studies examined purchasing using HEI [28,32,54]. Overall, higher SES households had significantly higher HEI total scores than lower SES households. This finding was not consistent across SES groups for certain HEI sub-component scores. Five studies examined kilocalories with four studies identifying significant differences cross-sectionally or overtime across groups [30,34,42,49,52]. However, the variety of outcome measures used in these studies make synthesis of the results difficult. Two studies found that purchases among lower SES households were higher in energy density and total calories per person per day compared to higher SES households [34,49]. Another study found that low-income households purchased fewer calories per capita per day, but also had the slowest decrease in calories purchased per capita per day over time (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) [42].
Among the four papers that examined nutrient purchasing outcomes [34,42,49,52], all four studies examined sugar, two examined saturated fat [49,52], and three studied sodium content [34,49,52]. There was a consistent pattern that lower SES groups had higher purchases of sugar than higher SES groups regardless of measures (total sugar, added sugar, sugar density).
Similarly, in the three studies that assessed sodium content of purchases, there was a consistent pattern that lower SES groups had higher purchases of sodium than higher SES groups. In the two studies that examined saturated fat content of purchases, one study did not find significant differences across SES groups [49], and the other found that the saturated fat content of purchases varied by store type for different SES groups [52].
Three studies examined purchasing outcomes that were not part of our primary outcomes of interest, so we classified the results of these studies into the Other category [29,50,51]. In two of these studies, SES was operationalized by using three groups (low, middle and high income groups) and both studies found significant differences across SES groups. In one study, high and middle income households had significantly higher proportions of purchases with pricing promotions compared to low-income households [51]. In the other study, high and middle income households had significantly higher proportions of purchases with low-content nutrient claims compared to low-income households [50]. The third study found no significant differences in grocery dollar spending on food and beverages among differing levels of education [29].

Single Factor Targeted Populations
Key findings from targeted studies are provided in Table S3 below. Six studies reported purchasing for a single factor targeted population [21,22,24,25,28,41]. These populations were low-income individuals or households [21,41] and individuals or households residing in an urban city [22,24,25,28]. Studies with a low-income targeted population focused on participants of federal food assistance programs such as SNAP and the Special Supplemental (2019) reported mean HEI-2010 total scores of 59 among customer purchases (from all stores) [22,25],  found a mean score of 39 among urban customer purchases from specifically limited-services (a score of 50 is considered average quality) [25]. Only two studies assessed kilocalorie, saturated fat, sugar, and/or sodium content [22,24]. Again, both studies reported high volumes among customer purchases. Page 22  X X X X X X X About 2%, 6%, 17%, and 8% of participants purchased 1 or more servings of fruit, vegetables, savory snacks, and artificially sweetened beverages, respectively. No differences were detected by store type. About 15%, 13% and 46% of participants reported purchasing 1 or more servings of candy, sweet baked goods, and SSB, respectively. Significant differences detected by store type. Average HEI-2010 total score for purchases was 36.4. No significant difference detected by store type. Overall, median of kcals purchased was 540 (253-1287) and median % of kcal from saturated fatty acids was 6.2 (0-13). Significant differences were detected by store type. Caspi (2017) [2] Urban (Residents of an urban city) X X X Among purchases, 8% included at least one fruit or vegetable. Increased amount (in pounds) of FV available in store and increase varieties of FV were significantly associated with greater odds of purchasing a fruit or vegetable. Among purchases, 8% included at least one serving of whole grains (most were snack items such as popcorn and tortilla chips). Pounds and varieties not associated with whole grain purchasing. Mean HEI-2010 total score for food purchases was 31 (±13). More store shelf space for fruits and beverages, higher healthy vs. unhealthy food ratio, and higher healthy food availability scores were associated with higher HEI-2010 purchase scores.
(Residents of lowincome community in an urban city) X X X X X Of total purchases, 5% of intercepts included fruit and 0% included vegetables. 21.7% included snack foods (no specification), 13.3% included candy, and 1.7% included ice cream. 55% of intercepts included beverages (no specification between sweetened and unsweetened)