1. Introduction
African Americans (AA) are at greater risk for developing chronic diseases compared to White (W) Americans [
1]. Socioeconomic status and lifestyle factors, especially diet, contribute to these health disparities. In fact, diet is considered one of the strongest modulators of chronic inflammation [
2,
3], a condition associated with the development of obesity, cardiovascular disease, cancer and diabetes [
4,
5,
6]. It has been estimated that half of all cardiovascular events could be prevented by improving the diet [
7].
Diets are composed of foods and beverages which contain not only macro- and micronutrients but also phytochemicals. Some of these compounds, such as carotenoids and polyphenols, may promote health and prevent disease [
8,
9]. Flavonoids are the most abundant polyphenols in the diet, accounting for approximately two-thirds of intake [
10]. Based on their chemical structure, they may be categorized into six classes: anthocyanidins, flavan-3-ols, flavanones, flavones, flavonols, and isoflavones [
11].
In vitro, the health-promoting actions of flavonoids are well documented. They have been shown to exhibit anti-inflammatory and antioxidant properties [
12,
13,
14]. It is apparent that these actions have implications for a variety of chronic diseases. Indeed, extensive research has found inverse associations between flavonoid intake and chronic diseases including cardiovascular disease [
7,
15,
16,
17,
18], cancer [
19,
20,
21,
22,
23], diabetes [
24,
25,
26,
27], and obesity [
28,
29,
30].
Despite their purported health benefits, knowledge about the flavonoid intake of at-risk populations is limited. In most large-scale investigations of flavonoid-health associations, the study sample was drawn from fairly homogenous populations. For example, in the Nurse’s Health Study I and II [
31], and the Health Professional’s Follow-up Study [
32], 97% of study participants were white and all were college educated [
33,
34]. One notable exception is the REGARDS (Reasons for Geographic and Racial Differences in Stroke) study, a national prospective cohort of community dwelling adults, which included a large proportion (42%) of black Americans [
35,
36]. Inverse associations between anthocyanidin intake and incident coronary heart disease [
35] and flavanone intake and incident ischemic stroke in both whites and blacks [
36] were reported. However, the design of the REGARDS study did not consider socioeconomic status when selecting the sample, nor were investigations of flavonoid intake-disease associations stratified by income. In addition, since REGARDS was focused on associations with stroke incidence, much of the sample was selected in the “Stroke Belt”, which encompasses the southern United States (U.S.) and includes rural, suburban, and urban communities [
35,
36,
37].
The Healthy Aging in Neighborhoods of Diversity across the Life Span (HANDLS) study was designed specifically to measure the associations between race and socioeconomic status, both independently and synergistically, and health disparities [
38]. Previously published research indicates the prevalence of risky health behaviors, including a nutrient-poor diet, is high among participants in HANDLS. For instance, Kuczmarski et al. found that most HANDLS participants consume a pro-inflammatory Westernized diet [
39]. The Westernized diet is characterized by high intakes of saturated fat, proteins (derived from processed meats), and sugars and low intakes of dietary fiber [
40]. Though it is assumed to be far less than optimal, flavonoid intakes, and thus the potential health benefits they could confer, are not known for this population. Further, whether these intakes differ by race and income is similarly not determined.
The objectives of this study were: (1) estimate flavonoid intakes of a large, racially and socioeconomically diverse urban population; (2) characterize major food and/or beverage contributors to intakes of total flavonoids and six flavonoid classes by race and income; (3) identify differences in intake between AA and W; and (4) identify differences in intake between higher and lower-income groups. Finally, because there is some evidence that AA have lower flavonoid intakes than W [
41], flavonoid intake estimates in the HANDLS study will be compared, for all and by race, to those of a nationally representative sample that is similar with regards to race and income. The purpose of this contrast is to ascertain if intakes observed in HANDLS were unique to this at-risk population or simply a reflection of intakes observed for the U.S. population as a whole.
4. Discussion
In this study, factors that are typically predictive of nutritional status and thus likelihood of nutrition-related disease were investigated in relation to their associations with flavonoid intake. These findings provide the first evidence that race, regardless of other potential mediating variables including sex, age, and income, was associated with differences in flavonoid intake in a large, at-risk population. With the sole exception of the flavanone class, intakes of flavonoids among AA were significantly lower than those of W. Conversely, income was not related to flavonoid intake.
It is difficult to incorporate these results with existing literature because flavonoid intake has not been investigated by race and income within other relevant characteristics. However, a few studies have analyzed flavonoid intake by race and income individually, and thus can be compared to our overall findings. In an investigation of flavonoid-incident ischemic stroke associations in REGARDS, Goetz and colleagues reported that AA had higher flavanone intakes as compared to W, while intake of total flavonoids and the other flavonoid classes were higher among W, though it is not clear if statistical testing was conducted [
36]. Analyses of WWEIA, NHANES data have also found racial differences in flavonoid intake. Sebastian et al. reported that non-Hispanic blacks had lower total flavonoid intake as compared to non-Hispanic whites or Hispanics [
41]. Differences in flavonoid class intake by race in earlier NHANES cycles have also been reported [
52], though which groups differed significantly in intake from others were not elucidated.
The relationship between flavonoid intake and income is less consistent. The majority of households included in the HANDLS study had incomes below
$75,000 [
53], which may have contributed to the observed lack of association with flavonoid intake. Financial resources could have been a limiting factor to achieving a diet replete in fruits and vegetables for all persons in the HANDLS population, and not just those in the lower-income category. This assertion is supported by earlier research that reported mean total HEI-2010 scores in baseline HANDLS participants were approximately 10 points lower than those of the U.S. population [
54]. It is well documented that lower quality diets cost less, and the foods they include are commonly selected by lower-income individuals [
55]. Nevertheless, these findings of no association concur with results from some studies, but not others. Using transformed flavonoid data from WWEIA, NHANES, which has a greater range of incomes than HANDLS, previously reported research found no differences in total flavonoid intake by income [
41]. However, other work analyzing earlier NHANES data cycles did report intake differences by income [
52,
56]. These later studies either did not transform the flavonoid data [
56] or used log transformation [
52], which proved inadequate to normalize the flavonoid distribution in the present study. In addition, those studies were not able to fully estimate flavonoid intake from all foods and beverages due to missing flavonoid composition data, which might account for the inconsistency with the current findings [
52,
56].
There may be many factors influencing the lower consumption of flavonoids by AA. Food selection decisions are complex, dependent on interactions between behavior and biology, and modulated by environment and community structure [
57]. Disparate intake of flavonoid classes by race could be a reflection of cultural food habits [
11,
58]. Previous research of HANDLS data has revealed food selections varied by race [
39,
54], which contributed to observed differences in dietary patterns with different 10-year atherosclerotic cardiovascular disease risk [
39]. When compared to AA, W had higher mean Healthy Eating Index (HEI)-2010 scores for the total vegetables and whole fruits components [
59], which would be consistent with W’s observed higher intakes of flavones and flavonols (both found primarily in vegetables) and anthocyanidins (found primarily in fruits, particularly berries and grapes). Similarly, it is clear that AA had higher intakes of the foods that are the primary contributors of flavanones in the American diet. Orange juice was the highest contributor of flavanone intake regardless of race. Post hoc analyses of HANDLS revealed that although the adjusted percentage of adults reporting orange juice did not differ by race (16 percent vs. 10 percent for AA and W, respectively;
p = 0.011), the adjusted mean amount consumed by AA (60 gm) was more than double that consumed by W (29 gm;
p < 0.01). Orange juice contains ~19 mg of flavanones per 100 gm [
50]. Based upon the intake estimates presented in this study, it can be concluded that orange juice intake alone accounted for approximately 80 percent of the difference in flavanone intake observed between AA and W.
Food availability is another factor that must be considered when disparities in intake are found. The concept of “food deserts”, defined as geographic areas with low availability or high prices of healthy foods, has been a popular one in recent years [
60]. It provides a rationale for why individuals who reside in areas with low healthful food availability may consume nutrient- and polyphenol-poor diets. However, its influence on dietary choice may be overstated [
60]. Instead, it may be that effort to obtain healthful foods [
61] (in addition to well-established economic considerations previously mentioned [
57]) is a more important determinant of fruit and vegetable and thus flavonoid intake than food availability [
57,
61], because travel outside one’s immediate neighborhood to acquire food is common [
57,
60]. This food shopping behavior is evident and impactful. Previous analyses of HANDLS revealed that, contrary to expectations, individuals who lived in neighborhoods with low healthy food availability had better diet quality than did their counterparts who resided in neighborhoods with medium and high healthy food availability [
61], suggesting that those in the former group traveled to obtain food. In another study also among an urban population, shopping at convenience and neighborhood stores, which are prevalent in the immediate environment, was positively associated with intake of added sugars and discretionary fat. Conversely, shopping at the types of stores that are typically not present in lower-income urban areas, such as specialty grocery stores and full-service supermarkets, was positively associated with fruit and vegetable intake [
62]. Furthermore, urban W adults appear to be more likely to travel to shop as compared to AA adults [
61]. The fact that W engage in this behavior to a greater extent than AA and thus potentially have more healthful choices available could explain their higher intake of total flavonoids and most flavonoid classes analyzed. This behavior is an area that deserves further study.
Other attributes of flavonoid-rich foods could potentially be important with respect to the differences observed in flavonoid intake between the exclusively urban HANDLS population and the comparable NHANES population. Anthocyanidin intakes in HANDLS were lower overall and by race relative to intakes in WWEIA, NHANES, due in part to the greater percentage in HANDLS with no intake of anthocyanidins at all (54.4 ± 1.2% vs 39.6 ± 1.9% in NHANES; unpublished NHANES data). Berries and grapes are primary sources of anthocyanidins in HANDLS and NHANES [
41]. Both are relatively expensive and perishable fruits, which are qualities that may discourage their purchase in a population who, besides having limited financial resources, may have to travel some distance to obtain them.
When flavonoid intake estimates in HANDLS and NHANES were compared within race, differences emerged. Among AA/blacks, for those flavonoid classes that differed between the surveys, intakes were consistently lower in the HANDLS population. In contrast, among W, intakes of anthocyanidins were lower in HANDLS but intake of total flavonoids and the flavan-3-ol and isoflavones classes were higher. These types of results exemplify the need to look at subpopulations individually, because disparate associations could lead to a conclusion of no relation for the population as a whole.
Overall, it may be that the unique characteristics of HANDLS play an important role in accounting for differences in flavonoid intake. The HANDLS sample, by design, targeted an at-risk population. As such, factors typically associated with health and health disparities beyond race and income are more pronounced in HANDLS as compared to the U.S. population as a whole. For instance, in 2005, among adults 25–64 years in the U.S., 12% did not have a high school degree or its equivalent [
63]. In the HANDLS sample, fully one-third fit this criterion. Education has been shown to be positively associated with intake of multiple Healthy Eating Index fruit and vegetable components–total vegetables, greens and beans, total fruit, and whole fruit [
54], all of which are positively associated with flavonoid intake [
41]. In addition, as employment status is highly correlated with education, it is not surprising that the percentage employed in the HANDLS sample (~56%) appears lower than 2005 national estimates (75.5%) [
64]. Unemployment status is another characteristic that has been shown to be negatively associated with fruit and vegetable consumption [
65]. It appears that lower intake of flavonoids is yet another marker of compromised dietary intake that places the population represented by HANDLS at greater nutritional risk [
39,
66,
67].
As with any research there are strengths and limitations. One major strength is that the results were based on analyses of data that was transformed to meet the assumptions of the statistical tests. Another strength is the relatively large sample of African Americans, a feature that is atypical of most studies investigating flavonoid intake. The flavonoid composition data and dietary data analyzed are other assets of this study. Use of the Flavonoid Database for USDA Food Codes permitted comprehensive estimation of flavonoid intake with minimal imputation to address missing flavonoid profiles for foods. The dietary collection method used in both HANDLS and WWEIA, NHANES, the USDA AMPM, has been shown to reduce bias in the collection of energy intakes, and to provide accurate estimates of sodium intake [
46,
68]. Nevertheless, inherent errors are associated with the 24-h recall [
69]. Furthermore, there are a plethora of other issues to consider with the use of self-reported dietary data that extends beyond the dietary assessment method. Social desirability can affect reporting of behaviors such as food and beverage intake. However, there seems to be no consistent pattern of either under- or over-reporting of healthful foods like fruits and vegetables [
70], so flavonoid estimation is likely minimally affected. Lastly, this analysis of one-day 24-h intakes per sample person does not address the long-term consumption of flavonoids. Rather, it provides a picture of daily mean intake of flavonoids for the HANDLS population on any given day of the year.