Fruit and Vegetable Intake Among Rural Populations in the United States, by Sociodemographic Characteristics, Behavioral Risk Factor Surveillance System 2019, 2021

Abstract

Background

Rural populations in the US experience a higher prevalence of chronic diseases compared with urban populations. Consumption of healthy foods in rural areas can be limited by factors such as reduced access and availability. Limited intake of fruits and vegetables is a common risk factor for chronic disease, but differences in intake by sociodemographic characteristics among rural populations is inadequately characterized.

Aim

We described the prevalence of consuming fruits and vegetables at least once per day by sociodemographic subgroups among US adults in rural counties.

Methods

We conducted a pooled, cross-sectional study of 218,905 US adults aged ≥18 years in rural counties (non-metropolitan and non-core counties defined by the National Center for Health Statistics) using 2019 and 2021 Behavioral Risk Factor Surveillance System data. Sociodemographic factors included age, sex, race or ethnicity, education, employment, income, healthcare access, US Census division, and state. The two primary outcomes were self-reported consumption of fruits and of vegetables at least once per day. Weighted prevalence estimates were calculated using predictive margins derived from adjusted logistic regression models.

Results

The prevalence of consuming fruits at least once per day was 57.6% (95% CI: [57.1-58.0]) and for vegetables was 80.0% (95% CI: [79.6-80.4]). For both fruits and vegetables, prevalence was generally higher among rural adults who were older; female; reported higher education, higher income, greater healthcare access; or resided in states in New England.

Conclusion

Fruit and vegetable consumption is inadequate across the population with some groups more likely to have lower consumption, such as younger adults, men, and adults of lower socioeconomic status. Interventions that address both population-level contextual factors and individual-level barriers for those most at risk for lower consumption may increase fruit and vegetable consumption.

Introduction

Compared with urban populations, rural populations in the US experience a higher risk of age-adjusted mortality (Garcia et al., 2019) and chronic diseases, such as obesity, diabetes, and cardiovascular disease (Coughlin et al., 2019). Adequate consumption of fruits and vegetables is essential for a healthy diet, and low consumption of fruits or vegetables can contribute to an increased risk for several of the most prevalent chronic diseases in the US including obesity, cardiovascular disease, type 2 diabetes, and some cancers (Stanaway et al., 2022; Yip et al., 2019) as well as worsened progression of existing chronic disease (Mąkosza et al., 2024; Jiménez-Cortegana et al., 2021). Access to fruits and vegetables can be a particular challenge in rural areas due to factors such as affordability and/or financial insecurity, low availability and quality of fruits and vegetables in local food outlets, and lack of transportation (Bardenhagen et al., 2017; Byker Shanks et al., 2022; Hendrickson et al., 2006). Improving and achieving adequate fruit and vegetable intake can be a practical intervention to help prevent and manage chronic disease, including in rural populations.

As part of a healthy dietary pattern, the 2020–2025 Dietary Guidelines for Americans (DGA) provides recommendations for adequate fruit and vegetable intake, recommending for individuals 14 years and over 1.5 to 2.5 cup-equivalents per day of fruits and 2 to 4 cup-equivalents per day for vegetables, depending on age, sex, and energy intake (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020). Studies show inadequate consumption of fruits and vegetables nationwide, with even lower intake of fruits and vegetables among adults in rural compared with urban areas (Lutfiyya et al., 2012). A previous study using survey data from 2019 in the general US population reported a median daily intake of once per day for fruits and 1.6 times per day for vegetables, corresponding to 12.3% of the population meeting DGA recommendations for fruits and 10.0% for vegetables (Lee et al., 2022). Another study, using survey data from 2009, examined adults living in rural areas and reported a greater frequency of fruit and vegetable intake among certain sociodemographic subgroups, such as adults who were female compared with male and those with a reported annual income of ≥$35,000 compared with below $35,000 (Lutfiyya et al., 2012). Other studies in rural populations are more limited in scope, evaluating fruit and/or vegetable intake in specific subpopulations or regions of the US (Cohen et al., 2018; Boeckner et al., 2007).

While these studies provide valuable insight into fruit and vegetable consumption in rural populations, it is also important to evaluate what sociodemographic factors are associated with low fruit and vegetable intake to facilitate tailored interventions. Demonstrating what factors in rural populations are most strongly associated with low intake of fruits and vegetables can inform public health practitioners and policy makers to help identify what types of interventions and which groups might benefit most from interventions that increase access and availability of healthier foods in rural areas. Therefore, we aimed to describe the prevalence of at least once per day fruit and vegetable intake among adults residing in rural areas in the US by sociodemographic subgroups.

Methods

Study Sample

This study was a cross-sectional analysis of pooled data from the 2019 and 2021 cycles of the Behavioral Risk Factor Surveillance System (BRFSS), an annual random-digit-dialed landline and cellular telephone-based survey of a randomly selected representative sample of noninstitutionalized US adults in each state. Respondents are aged ≥18 years, from the 50 US states, District of Columbia, and participating territories (Guam, Puerto Rico, US Virgin Islands) (Centers for Disease Control and Prevention. National Center for Chronic Disease Prevention and Health Promotion. Division of Population Health). BRFSS collects self-reported information from respondents on sociodemographic characteristics, health-related risk factors, healthcare access, and comorbidities. Telephone interviewers obtain verbal informed consent by reading a script and describing the survey and confidentiality of collected information. The present analysis used BRFSS data from 2019 and 2021, the two most recent surveys with data available on fruit and vegetable intake. The analysis included adults who resided in a county classified as micropolitan or non-core according to the National Center for Health Statistics’ Urban-Rural Classification Scheme for Counties (Centers for Disease Control and Prevention. National Center for Health Statistics). This classification excluded Connecticut, Delaware, the District of Columbia, Rhode Island, and New Jersey as these states did not have any micropolitan or non-core counties. Results for Florida were based on survey data from 2019 as no data were available for the 2021 survey. Institutional review board approval was not required as the study comprised analysis of secondary, deidentified data.

Measurements

The questionnaire included in total six questions asking respondents how often they consumed fruits and vegetables during the past 30 days (Supplemental Table 1). Questions were adapted from a validated module from the National Cancer Institute’s Dietary Screener Questionnaire and collected information on consumption of: fruits (excluding juices), 100% fruit juice, green leafy vegetables, fried potatoes, other potatoes, and other vegetables. Responses could be provided in times per day, per week, or per month. These responses were converted into the primary outcomes of self-reported consumption of: 1) fruits at least once per day; 2) vegetables at least once per day. Consistent with prior literature (Lee et al., 2022), fruit consumption included whole fruit and 100% fruit juice, while vegetable consumption included green leafy vegetables, fried potatoes, other potatoes, and other vegetables. Although recommended fruit and vegetable consumption is based on cup-equivalents and varies by age, sex, and total caloric intake (U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2020), the survey questions did not collect information on the amount of fruits and vegetables consumed such as cup-equivalents. Therefore, we used daily consumption, as regardless of the differing recommended amounts, all persons should consume fruits and vegetables daily.

Sociodemographic variables included age (18–34, 35–44, 45–64, ≥65 years), sex (female, male), education (less than high school, high school, more than high school), employment status (employed, not employed, retired) and annual household income (<$15,000, $15,000–24,999, $25,000–34,999, $35,000–49,999, $50,000–74,999, ≥$75,000). Race and ethnicity categories included: Hispanic, non-Hispanic (NH) American Indian/Alaska Native, NH Asian, NH Black, NH Pacific Islander, NH White, NH Multiracial, NH Other, Missing. Geographic variables included US Census Division (Pacific, Mountain, West North Central, West South Central, East North Central, East South Central, South Atlantic, Middle Atlantic, New England) (United States Census Bureau) and state. Variables related to healthcare access included: 1) healthcare cost concerns, based on the respondent reporting at least one instance in the past 12 months of not seeing a doctor due to cost, and 2) no routine physical exam in the past 12 months.

Among the 856,958 adults in the 2019 and 2021 BRFSS survey cycles, 583,021 were excluded due to residing in metropolitan (i.e. non-rural) counties and 15,512 were excluded due to residing in US territories. Among the resulting 258,425 adults residing in rural counties, 39,520 adults had missing or extreme values for fruit or vegetable intake (extreme values were defined a priori by BRFSS as a self-reported intake of >16 times per day for fruits or >14 times per day for vegetables), resulting in an analytic cohort of 218,905 adults. Within this analytic cohort, any variable with ≥1% missing data (race or ethnicity [n=3,826; 1.7%], income [n=35,926; 16.4%]) used a missing indicator variable. For any variable with <1% missing data (education [n=545; 0.3%], employment status [n=1,232; 0.6%], healthcare cost concerns [n=471; 0.2%], no physical exam in the past 12 months [n=2,363; 1.1%]), respondents who had missing data were excluded from any analyses requiring the given variable.

Statistical Analysis

The primary analyses comprised adjusted logistic regression models. Weighted prevalence estimates and corresponding 95% confidence intervals (CI) were calculated using the PREDMARG function in SUDAAN software. The PREDMARG function estimates the predictive margins, which represent the sample weighted average of the predicted response (i.e. prevalence of consumption of fruits or vegetables at least once per day) of a given group after adjusting for model covariates (Graubard and Korn, 1999; Bieler et al., 2010). We used two models for prevalence estimates. Model 1 adjusted for age, and Model 2 adjusted for age, sex, race or ethnicity, and education. Model covariates were selected a priori as known or plausible confounders. A covariate was excluded from a model if also used as a stratifying variable. For each model, sensitivity analyses were conducted, excluding 100% fruit juices in the estimation of fruit consumption, and excluding fried potatoes for vegetable consumption. For prevalence estimates by state, cut-offs for categories were chosen a priori to provide an approximately equal statistical distribution across categories. Sample weights and design variables accounted for the complex survey design. We considered analyses statistically significant if they had a two-sided p-value <0.05. All presented estimates met criteria for statistical reliability (had a relative standard error <0.30 and were produced from ≥50 unweighted records). We used SAS 9.4 (SAS Institute, Cary, North Carolina) for data management and SUDAAN 11.0.4 (RTI International, Research Triangle Park, North Carolina) for biostatistical analyses.

Results

Study Population

Table 1 shows characteristics of the 218,905 US adults residing in rural counties. Overall, the age group with the largest proportion of adults was 45–64 years (33.9% [95% CI: 33.5–34.3], and 48.7% (95% CI: 48.2–49.2) were female. Slightly over half of adults reported greater than a high school education (52.4% [95% CI: 51.9–52.9]) and current employment (53.3%, 95% CI: [52.8–53.8]). The annual household income level with the largest proportion of respondents was ≥$75,000 (24.1% [95% CI: 23.7–24.5]). Adults reporting healthcare cost concerns within the past 12 months comprised 11.6% (95% CI: 11.3–12.0) of adults, and 76.0% (95% CI: 75.6–76.4) reported having a physical examination in the past 12 months.

Table 1.

Characteristics of Adults in US Rural Areas, Behavioral Risk Factor Surveillance System, 2019 and 2021 (n=218,905)^a

Characteristic	Weighted Prevalence (95% confidence interval)	Unweighted n
Age, years
18–34	25.3 (24.8-25.8)	27,798
35–44	14.8 (14.5-15.1)	24,442
45–64	33.9 (33.5-34.3)	77,768
≥65	26.0 (25.6-26.4)	88,897
Sex
Female	48.7 (48.2-49.2)	98,425
Male	51.3 (50.8-51.8)	120,480
Race or Ethnicity
Hispanic	7.4 (7.0-7.7)	9,262
NH AIAN	2.0 (1.9-2.1)	6,542
NH Asian	0.9 (0.8-1.0)	1,799
NH Black	6.6 (6.4-6.9)	7,993
NH Pacific Islander	0.1 (0.1-0.1)	546
NH White	79.7 (79.2-80.1)	183,366
NH Multiracial	1.3 (1.3-1.4)	4,338
NH Other	0.4 (0.3-0.4)	1,233
Missing	1.7 (1.5-1.8)	3,826
Education
Less than high school	13.1 (12.7-13.5)	14,175
High school	34.5 (34.0-34.9)	66,033
More than high school	52.4 (51.9-52.9)	138,152
Employed
Yes	53.3 (52.8-53.8)	105,372
No	23.7 (23.3-24.1)	38,001
Retired	23.0 (22.7-23.3)	74,300
Annual household income
<$15,000	7.8 (7.6-8.1)	15,312
$15,000–24,999	12.9 (12.6-13.3)	26,516
$25,000–34,999	11.0 (10.8-11.3)	23,602
$35,000–49,999	12.8 (12.5-13.1)	28,827
$50,000–74,999	14.4 (14.1-14.7)	32,980
≥$75,000	24.1 (23.7-24.5)	55,742
Missing	16.9 (16.5-17.2)	35,926
US Census Division
Pacific	6.7 (6.5-7.0)	15,032
Mountain	8.7 (8.5-8.8)	33,923
West North Central	14.2 (14.0-14.4)	61,467
West South Central	12.2 (11.9-12.6)	11,563
East North Central	19.1 (18.8-19.5)	20,578
East South Central	13.4 (13.2-13.6)	17,150
South Atlantic	15.7 (15.4-16.0)	21,549
Middle Atlantic	6.4 (6.2-6.7)	11,916
New England	3.5 (3.5-3.6)	25,727
Healthcare cost concerns in past 12 monthsb	11.6 (11.3-12.0)	18,648
Physical exam in past 12 months	76.0 (75.6-76.4)	172,399

Abbreviations: AIAN: American Indian or Alaska Native; NH = Non-Hispanic

^aRural areas were defined as counties classified as micropolitan or non-core according to the National Center for Health Statistics’ Urban-Rural Classification Scheme for Counties

^bReported a time in the past 12 months when a doctor could not be seen due to cost.

Fruit Intake

Table 2 shows the prevalence of adults in rural areas reporting consumption of fruits at least once per day by sociodemographic subgroups. Overall, 57.6% (95% CI: 57.1–58.0) of adults reported consuming fruits at least once per day. Across subgroups, results differed little between Model 1 (adjusted for age) and Model 2 (further adjusted for sex, race, and education). In Model 2, female adults (59.8%, 95% CI: [59.2–60.4]) had a higher prevalence compared with male adults (55.2, 95% CI: [54.5–55.9]). Among race or ethnicity groups, lowest prevalence estimates were among NH White (56.6% [95% CI: 56.1–57.1]) and NH AIAN adults (56.8% [95% CI: 54.3–59.3]) and the highest prevalence was among Hispanic (65.2% [95% CI: 62.8–67.5]) and NH Other adults (69.1% [95% CI: 62.1–75.3]). Markers of higher socioeconomic status, such as having more than a high school education or higher annual household income, were associated with higher prevalence estimates for fruit consumption at least once per day. For example, among different annual household income levels, the highest prevalence was among adults reporting an income of ≥$75,000 (61.2% [95% CI: 60.2–62.1]). Among US Census Divisions, prevalence ranged from 53.5% (95% CI: 52.4–54.6) in the East South Central division to 65.6% (95% CI: 64.5–66.7) in New England. Markers of greater healthcare access were associated with higher prevalence of consuming fruits at least once per day. For example, adults who did not report any healthcare cost concerns in the past 12 months had a prevalence of 58.2% (95% CI: 57.7-58.8), compared with those who did report healthcare cost concerns (53.0% [95% CI: 51.5–54.5]). In sensitivity analyses excluding 100% fruit juices (Supplemental Table 2), the prevalence of adults who reported consuming fruits at least once per day overall was slightly lower at 49.3% (95% CI: 48.8-49.7), and differences among subgroups showed a similar pattern compared with prevalence estimates that included 100% fruit juices.

Table 2.

Percent of Adults in US Rural Areas Reporting at Least Once Per Day Consumption of Fruits, by Characteristic, BRFSS 2019 and 2021^a

Characteristic	Prevalence (95% Confidence Interval)
	Model 1b	p-valuec	Model 2d	p-valuec
Overalle	57.6 (57.1–58.0)	–	57.6 (57.1–58.0)	–
Age, years
18–34	55.6 (54.5–56.8)	<0.001	55.4 (54.3–56.5)	<0.001
35–44	56.8 (55.6–58.0)		56.3 (55.0–57.5)
45–64	54.8 (54.1–55.6)		55.0 (54.3–55.7)
≥65	63.4 (62.7–64.1)		63.8 (63.0–64.5)
Sex
Female	60.0 (59.4–60.6)	<0.001	59.8 (59.2–60.4)	<0.001
Male	55.0 (54.3–55.6)		55.2 (54.5–55.9)
Race or Ethnicity
Hispanic	63.7 (61.3–66.1)	<0.001	65.2 (62.8–67.5)	<0.001
NH AIAN	56.2 (53.7–58.6)		56.8 (54.3–59.3)
NH Asian	62.0 (56.7–67.1)		60.9 (55.5–66.0)
NH Black	58.6 (56.6–60.5)		59.1 (57.1–61.0)
NH Pacific Islander	57.7 (51.6–63.6)		58.1 (51.9–64.0)
NH White	56.8 (56.3–57.3)		56.6 (56.1–57.1)
NH Multiracial	58.2 (55.0–61.4)		58.4 (55.2–61.6)
NH Other	68.6 (61.3–75.2)		69.1 (62.1–75.3)
Missing	60.0 (56.7–63.1)		60.3 (57.0–63.5)
Education
Less than high school	54.3 (52.7–55.8)	<0.001	53.4 (51.8–54.9)	<0.001
High school	54.7 (53.8–55.5)		54.9 (54.0–55.7)
More than high school	60.3 (59.7–60.9)		60.4 (59.8–61.0)
Employed
Yes	57.4 (56.7–58.0)	0.001	57.4 (56.7–58.0)	0.001
No	56.4 (55.3–57.4)		56.3 (55.3–57.4)
Retired	59.2 (58.1–60.3)		59.2 (58.1–60.3)
Annual household income
<$15,000	52.9 (51.1-54.6)	<0.001	52.8 (51.0-54.5)	<0.001
$15,000–24,999	54.6 (53.3-55.9)		54.5 (53.2-55.8)
$25,000–34,999	56.6 (55.2-58.0)		56.6 (55.2-58.0)
$35,000–49,999	56.7 (55.5-58.0)		57.0 (55.8-58.2)
$50,000–74,999	59.2 (58.0-60.4)		59.3 (58.1-60.5)
≥$75,000	61.2 (60.3-62.1)		61.2 (60.2-62.1)
Missing	56.7 (55.5-57.8)		56.5 (55.3-57.6)
US Census Division
Pacific	62.6 (60.5-64.6)	<0.001	61.8 (59.6-63.9)	<0.001
Mountain	59.7 (58.9-60.5)		58.6 (57.8-59.5)
West North Central	57.5 (56.9-58.2)		57.9 (57.2-58.5)
West South Central	54.4 (52.4-56.4)		53.5 (51.6-55.4)
East North Central	58.1 (57.1-59.2)		58.9 (57.8-60.0)
East South Central	53.2 (52.1-54.3)		53.5 (52.4-54.6)
South Atlantic	56.5 (55.3-57.7)		56.3 (55.1-57.5)
Middle Atlantic	61.3 (59.2-63.3)		62.1 (60.1-64.1)
New England	65.4 (64.3-66.5)		65.6 (64.5-66.7)
Healthcare cost concerns in past 12 months
No	58.2 (57.7-58.6)	<0.001	58.2 (57.7-58.7)	<0.001
Yes	53.0 (51.5-54.6)		53.0 (51.5-54.5)
Physical exam in past 12 months
No	54.3 (53.3-55.3)	<0.001	54.9 (53.9-55.9)	<0.001
Yes	58.7 (58.1-59.2)		58.5 (57.9-59.0)

Abbreviations: AIAN: American Indian or Alaska Native; BRFSS: Behavioral Risk Factor Surveillance System; NH = Non-Hispanic

Covariates are excluded from a model if also used as a stratifying variable.

^aRural areas were defined as counties classified as micropolitan or non-core according to the National Center for Health Statistics’ Urban-Rural Classification Scheme for Counties

^bModel 1 covariates include: age

^cP-values represent a group test for overall effect

^dModel 2 covariates include: age, sex, race or ethnicity, education

^eOverall estimates derived from predicted margins in both models are mathematically equivalent (i.e. unadjusted)

Figure 1 shows the prevalence of adults in rural areas reporting consumption of fruits at least once per day by state. Prevalence estimates were highest (≥62.5%) in California, Colorado, Maine, Massachusetts, New Hampshire, New York, Utah and Vermont, followed by Minnesota, Montana, Oregon, Pennsylvania, South Dakota, Washington, and Wisconsin at 60.0–62.4%. Eleven states had a prevalence of 57.5-59.9%, eight states a prevalence of 55.0-57.4%, and seven states a prevalence of 52.5-54.9%. The remaining five states (Alabama, Arkansas, Louisiana, Mississippi, Oklahoma) had the lowest prevalence estimates of <52.5%.

Figure 1. Percent of Adults in US Rural Areas Reporting at Least Once Per Day Consumption of Fruits, by State, BRFSS 2019 and 2021.

Areas in grey are those with no rural counties (Connecticut, Delaware, District of Columbia, New Jersey, Rhode Island). Estimates for Florida are based on survey data from 2019 as no data were available for the 2021 survey. Estimates are derived from Model 2, which adjusted for age, sex, race or ethnicity, and education.

Vegetable Intake

Prevalence of at least once per day consumption of vegetables in the overall population was 80.0% (95% CI: 79.6–80.4) (Table 3). Similar to estimates for fruit intake, prevalence estimates within subgroups were similar across Model 1 and Model 2. In Model 2, the highest and lowest prevalence among age groups was for adults aged 35–44 years (81.5% [95% CI: 80.5–82.4]) and those aged 18-34 years (78.3% [95% CI: 77.4-79.3]), respectively. Female adults (82.0% [95% CI: 81.4–82.4]) had a higher prevalence compared with male adults (78.0% [95% CI: 77.5–78.6]). Among race or ethnicity groups, prevalence ranged from 70.8% (95% CI: 69.0–72.5) among NH Black adults to 86.3% (95% CI: 81.8–89.9) among NH Other adults. Across US Census Divisions, prevalence ranged from 78.6% (95% CI: 78.0–79.1) in the West North Central division to 84.9% (95% CI: 84.0–85.8) in New England. Markers of greater socioeconomic status (higher education, current employment, higher income) were also associated with higher prevalence estimates. For example, for educational attainment, adults reporting more than a high school education had the highest prevalence (84.0% [95% CI: 83.5–84.4]). Across levels of annual household income, prevalence ranged from 73.1% (95% CI: 71.6–74.5) for adults with an income of <$15,000 to 84.7% [95% CI: 83.9–85.4] among adults with an income ≥$75,000. In sensitivity analyses that excluded fried potatoes from vegetable consumption (Supplemental Table 3), prevalence in the overall population was slightly lower at 71.2% (95% CI: 70.8-71.7). Differences by subgroups were similar when compared with results that included fried potatoes.

Table 3.

Percent of Adults in US Rural Areas Reporting at Least Once Per Day Consumption of Vegetables, by Characteristic, BRFSS 2019 and 2021^a

Characteristic	Prevalence (95% Confidence Interval)
	Model 1b	p-valuec	Model 2d	p-valuec
Overalle	80.0 (79.6–80.4)	–	80.0 (79.6–80.4)	–
Age, years
18–34	77.5 (76.5–78.5)	<0.001	78.3 (77.4–79.3)	<0.001
35–44	81.3 (80.4–82.3)		81.5 (80.5–82.4)
45–64	80.7 (80.1–81.3)		80.6 (80.0–81.2)
≥65	80.7 (80.1–81.3)		80.1 (79.4–80.7)
Sex
Female	82.2 (81.7–82.7)	<0.001	82.0 (81.4–82.4)	<0.001
Male	77.7 (77.1–78.3)		78.0 (77.5–78.6)
Race or Ethnicity
Hispanic	68.6 (66.2–71.0)	<0.001	71.7 (69.4–73.9)	<0.001
NH AIAN	80.0 (77.9–81.9)		81.0 (79.0–82.8)
NH Asian	81.4 (77.2–85.0)		80.1 (75.7–83.9)
NH Black	69.8 (67.9–71.5)		70.8 (69.0–72.5)
NH Pacific Islander	76.9 (71.4–81.6)		77.5 (72.0–82.1)
NH White	81.9 (81.5–82.2)		81.6 (81.2–82.0)
NH Multiracial	80.8 (78.3–83.1)		81.0 (78.6–83.2)
NH Other	85.9 (81.3–89.5)		86.3 (81.8–89.9)
Missing	80.5 (77.9–82.9)		80.5 (77.8–83.0)
Education
Less than high school	70.4 (69.0–71.8)	<0.001	72.4 (71.0–73.7)	<0.001
High school	77.0 (76.3–77.6)		77.2 (76.5–77.9)
More than high school	84.4 (84.0–84.9)		84.0 (83.5–84.4)
Employed
Yes	81.4 (80.9–81.9)	<0.001	81.2 (80.7–81.8)	0.002
No	75.6 (74.7–76.5)		76.8 (75.9–77.7)
Retired	81.6 (80.7–82.4)		81.0 (80.1–81.8)
Annual household income
<$15,000	70.2 (68.6–71.8)	<0.001	73.1 (71.6–74.5)	<0.001
$15,000–24,999	75.0 (73.8–76.1)		76.9 (75.8–78.0)
$25,000–34,999	79.0 (77.8–80.1)		79.9 (78.8–81.0)
$35,000–49,999	80.4 (79.3–81.4)		80.3 (79.2–81.3)
$50,000–74,999	83.8 (82.8–84.7)		82.9 (81.9–83.8)
≥$75,000	86.1 (85.4–86.8)		84.7 (83.9–85.4)
Missing	76.8 (75.8–77.8)		77.3 (76.3–78.2)
US Census Division
Pacific	81.7 (80.0–83.3)	<0.001	80.9 (79.1–82.6)	<0.001
Mountain	81.2 (80.6–81.8)		81.1 (80.4–81.8)
West North Central	79.7 (79.1–80.2)		78.6 (78.0–79.1)
West South Central	76.9 (75.1–78.6)		78.9 (77.3–80.4)
East North Central	80.3 (79.4–81.1)		79.2 (78.2–80.1)
East South Central	79.3 (78.5–80.2)		80.2 (79.3–81.0)
South Atlantic	80.2 (79.2–81.2)		81.5 (80.5–82.4)
Middle Atlantic	79.8 (78.0–81.5)		79.1 (77.3–80.8)
New England	86.6 (85.7–87.4)		84.9 (84.0–85.8)
Healthcare cost concerns in past 12 months
No	80.4 (80.0–80.8)	<0.001	80.2 (79.8–80.6)	0.06
Yes	77.2 (75.7–78.5)		79.0 (77.7–80.2)
Physical exam in past 12 months
No	79.0 (78.2–79.8)	0.003	79.7 (78.8–80.5)	0.19
Yes	80.5 (80.0–80.9)		80.3 (79.8–80.7)

Abbreviations: AIAN: American Indian or Alaska Native; BRFSS: Behavioral Risk Factor Surveillance System; NH = Non-Hispanic

Covariates are excluded from a model if also used as a stratifying variable.

^aRural areas were defined as counties classified as micropolitan or non-core according to the National Center for Health Statistics’ Urban-Rural Classification Scheme for Counties

^bModel 1 covariates include: age

^cP-values represent a group test for overall effect

^dModel 2 covariates include: age, sex, race or ethnicity, education

^eOverall estimates derived from predicted margins in both models are mathematically equivalent (i.e. unadjusted)

In Figure 2, states with the highest prevalence (≥85.0%) of at least once per day consumption of vegetables included Florida and Maine. States with a prevalence of 82.5–84.9% included Idaho, Massachusetts, New Hampshire, North Carolina, and Vermont. Sixteen states had a prevalence of 80.0-82.4%, and twenty states a prevalence of 77.5–79.9%. Three states (Iowa, Minnesota, North Dakota) had the lowest prevalence estimates of <77.5%.

Figure 2. Percent of Adults in US Rural Areas Reporting at Least Once Per Day Consumption of Vegetables, by State, BRFSS 2019 and 2021.

Areas in grey are those with no rural counties (Connecticut, Delaware, District of Columbia, New Jersey, Rhode Island). Estimates for Florida are based on survey data from 2019 as no data were available for the 2021 survey. Estimates are derived from Model 2, which adjusted for age, sex, race or ethnicity, and education.

Discussion

In a large nationwide sample of US adults living in rural counties, the majority of the population reported consuming fruits and consuming vegetables at least once per day. Statistically significant differences in prevalence estimates appeared across several sociodemographic subgroups with generally higher reported intakes among subgroups such as adults who were older, female, had higher education, higher income, or greater healthcare access. States in New England, in particular, showed a higher prevalence of at least once per day consumption of fruits and vegetables. Furthermore, once per day consumption of fruits and vegetables represented a minimal threshold for adequate intake, suggesting future improvement may be needed to increase fruit and vegetable consumption.

For demographic variables (age, sex) and markers of socioeconomic status (education, income), associations between these variables and fruit or vegetable intake were similar when compared with prior studies of US adults in rural areas (Lutfiyya et al., 2012; Johnson et al., 2010) and other studies assessing the general US population comprising both rural and non-rural areas (McCullough et al., 2022; Ansai and Wambogo, 2021). This consistency suggests that, at least based on the measured variables, much of the demographic and socioeconomic factors associated with fruit and vegetable intake in the general US population likely do not differ from those in rural areas. Prevalence estimates for at least once per day fruit and vegetable consumption among race or ethnicity groups varied for some racial and ethnic groups. Aside from NH Other adults, who comprise a heterogeneous group whose corresponding findings are difficult to interpret, Hispanic adults had the highest prevalence for fruit intake. Prior studies have reported similar findings (McCullough et al., 2022; Lutfiyya et al., 2012; Colón-Ramos et al., 2009), although it is unclear to what extent this finding might be attributed to actual differences in fruit intake from factors including cultural differences in dietary preference or methodological limitations such as differential misclassification in self-reported intake.

Differences in fruit and vegetable intake by age, sex, and other sociodemographic subgroups have been widely reported previously and are consistent with findings of the current study. The greater consumption of fruits and vegetables in older adults can be attributed at least in part to a greater awareness of healthy eating to prevent or manage hypertension, cardiovascular disease, diabetes, and other conditions that are more prevalent among older adults (Nicklett and Kadell, 2013). The greater consumption of fruits and vegetables among women compared to men has also been reported previously, and is likely attributable to psychosocial factors such as cultural norms and greater health consciousness among women (Feraco et al., 2024; Lee-Kwan et al., 2017; Emanuel et al., 2012). The observed socioeconomic gradient in fruit and vegetable consumption can be attributed to a variety of factors such as fewer economic constraints, greater access to fresh foods, and greater health literacy among adults of higher socioeconomic status (Drewnowski and Rehm, 2015; Darmon and Drewnowski, 2015; Kell et al., 2015).

Markers of greater healthcare access or utilization were positively associated with higher prevalence estimates for fruit and vegetable intake, although associations attenuated slightly after further adjustment for sex, race or ethnicity, and education, and were no longer statistically significant for vegetable intake. Although we did not directly study any longitudinal association between healthcare access or utilization and fruit and vegetable intake, we can hypothesize as to possible explanations. For example, since the observed associations attenuated after further adjustment, socioeconomic factors associated with both healthcare access and utilization and fruit or vegetable intake, such as use of public insurance or transportation barriers, might partly explain the findings. Moreover, healthcare itself might have an emerging role in increasing fruit and vegetable intake through facilitating access to interventions such as produce prescriptions, medically-tailored meals, and other food-based interventions (Mozaffarian et al., 2024). However, increasing the role of healthcare might present challenges in rural settings which already experience a lack of healthcare providers and facilities (Nielsen et al., 2017) and other resources such as provider training needed to implement and maintain programs to increase fruit and vegetable intake (Coward et al., 2021). Regardless, these findings highlight opportunities for healthcare practitioners to provide dietary guidance and services to adults in rural populations with the goal of increasing fruit and vegetable intake.

Differences in fruit and vegetable intake across states highlight the geographic heterogeneity within rural populations. For example, states typically categorized by other studies as part of the rural South, which generally had the lowest fruit intake, are characterized by higher rates of morbidity and mortality (Miller and Vasan, 2021), and a higher prevalence of risk factors such as less access to healthcare providers, lower income, being uninsured, and experiencing food insecurity, when compared with other rural populations (Meit et al., 2014; Randolph et al., 2023; Miller and Vasan, 2021). In contrast, states typically categorized as part of the rural Northeast, which had the highest fruit and vegetable intake, tend to have the lowest rates of morbidity and mortality and lower prevalence of residents being uninsured and experiencing food insecurity compared with other rural populations and the general US population (Miller and Vasan, 2021; Randolph et al., 2023). Therefore, rather than a one size-fits-all approach in rural populations, a greater impact on improving fruit and vegetable intakes might be made by tailoring interventions to address the unique barriers to fruit and vegetable intake specific to a given state.

Regardless of statistically significant differences in fruit or vegetable intake by sociodemographic subgroups, the frequency of consumption of fruits and vegetables was low across all subgroups given the modest threshold of fruit and vegetable intake at least once per day. These findings suggest that in addition to intervening on individual-level factors, other factors such as policy, systems, and environmental change strategies, or contextual population-level factors might also need to be addressed to improve fruit and vegetable intake in rural settings. For example, contextual factors that might be associated with increased fruit and vegetable intake include proximity to supermarkets, increased availability of healthier foods at food outlets, lower costs for healthier food options, increased transportation options and culturally appropriate messaging (Fernqvist et al., 2024; Zenk et al., 2009; Fergus et al., 2021; Holston et al., 2020). In terms of policy, systems, and environmental change strategies, multiple programs from CDC’s Division of Nutrition, Physical Activity and Obesity support state and local public health organizations to advance such strategies to increase healthy lifestyle behaviors such as healthy eating (Centers for Disease Control and Prevention, 2024). These nutrition strategies include use of food service and nutrition guidelines in community institutions and expanding fruit and vegetable voucher and produce prescription programs.

This study is subject to limitations. First, only select sociodemographic variables were included in the BRFSS survey questionnaire. As a result, other potentially important factors such as those related to the built environment, transportation, housing, or food assistance programs could not be assessed. Second, fruit and vegetable intake were self-reported. This may result in misclassification, such as overreporting intake due to social desirability bias (Miller et al., 2008), misclassification that differs by sex (Burrows et al., 2019), and/or recall bias (Willett, 2012). Third, the BRFSS questionnaire asked respondents about frequency of consumption pertaining to the past 30 days, which may not represent an individual’s usual long-term dietary intake and does not include the amount consumed. Fourth, the degree of missing data might affect the generalizability of the results to the nationwide population. Lastly, the study used observational data and therefore is subject to residual confounding.

Fruit and vegetable consumption is inadequate across the population with some groups more likely to have lower consumption than others. Both interventions that address contextual factors at the population level as well as those that address the specific barriers for those most at risk for lower consumption may be needed to improve fruit and vegetable consumption. Future opportunities lie in identifying interventions with the greatest potential public health impact that can be feasibly implemented and sustained in rural settings.