Skip to content

Advertisement

  • Research
  • Open Access

The intersection of food insecure populations in the Midwest U.S. and rates of chronic health conditions

  • 1,
  • 1Email author,
  • 2,
  • 1 and
  • 3
Agriculture & Food Security20187:43

https://doi.org/10.1186/s40066-018-0195-z

  • Received: 16 February 2017
  • Accepted: 18 June 2018
  • Published:

Abstract

Background

Food insecurity is the state of having insufficient access to adequate food in order to maintain a healthy lifestyle due to limited economic resources. This study expands upon the annual survey conducted by the USDA ERS, while providing evidence that additional factors, notably, medical or health-related issues play vital roles in the relationship between households and food security.

Methods

The data for this study were generated by surveying Midwestern residents. The sample of respondents was targeted to be representative of the Midwest in terms of sex, age, and income and was completed by 1265 respondents. The survey included the CPS Food Security Supplement to assess the food security of respondents. All respondents were asked the 10 household focused food security questions and respondents who indicated having children were asked the additional 8 child focused questions. Additionally, respondents were asked demographic, food security, and health status questions.

Results

Of the sample, 25% were considered food insecure. Being male, middle aged, having children, having household diabetes, having a household eating disorder, and having household depression/anxiety are significant determinants of decreased food security.

Conclusions

Establishing clear relationships between health and food can help to inform legislation. This analysis suggests the inclusion of chronic illness and health information to improve metrics and inform food security legislation.

Keywords

  • Food insecurity
  • Regional
  • Chronic health
  • Midwest

Background

During election years, media and news sources overflow with information about candidates and platforms, covering a spectrum of political, social, and economic issues. Forbes featured an article presenting data from Pew Research about the 2016 presidential campaign, which found that 84% of voters deemed the “economy” the leading ballot issue which was followed by “terrorism,” “foreign policy,” and “health care” [1]. One component of the legislature that impacts the economy and addresses various aspects of food-related issues is ratified between elections—the Farm Bill. In 2014, President Barack Obama signed the latest Farm Bill into effect and commented to an audience at Michigan State University, “the Farm Bill is not just about helping farmers,” and explained that it covers economic initiatives such as jobs, innovation, and infrastructure [2]. Importantly, the largest portion of the Farm Bill is dedicated to nutrition assistance programs. According to the United States Department of Agriculture’s (USDA) “Budget Summary and Annual Performance Plan: FY 2016,” expenditures to meet mandatory and discretionary Farm Bill programs for 2016 were estimated to be $148 billion and 73% of budget outlays were for nutrition assistance [3]. President Obama explained “this country has helped Americans put food on the table when they hit a rough patch, or when they’re working hard but aren’t making enough money to feed their kids” [2].

Food security and chronic health

One example of a “rough patch” President Obama included was household illness [2]. It is understood that food assistance programs, such as the Supplemental Nutrition Assistance Program (SNAP), are designed to be temporary solutions, covering short-term crisis by providing aid for a few months, of which illness is considered a temporary economic strain on households [2].

A number of studies have analyzed and discussed a relationship between food security and illness. Using the USDA Economic Research Service (ERS) 30-day measure (a derivative survey of the Current Populations Survey (CPS) Food Security Supplement), Knight et al. [4] found 17.0% of respondents with diabetes were food insecure, and food insecure individuals were more likely to be medically uninsured. Furthering health considerations, Knight et al. [4] found that 18.9% of respondents with diabetes scrimped on medication (delaying filling prescriptions, inability to afford medication, taking less medication) and scrimping was positively and strongly correlated with food insecurity.

A similar study, also using the USDA ERS 30-day food security measure, found that 22% of adults with chronic diseases were food insecure or had cost-related medication under use and 11% reported both [5]. Berkowitz et al. [5] also found that participants of Women Infant Children (WIC) programs had decreased instances of food insecurity/cost-related medication underuse and suggest investigating the program dynamics. Muldoon et al. [6] found that there were higher odds of mental illness (specifically depression and anxiety) among those who experience insufficient food and hunger compared to those without hunger. A study done by Dharamasen et al. [7] found correlations linking food insecurity to health, including positive correlations between food insecurity and adult obesity, adult obesity and poverty, and adult obesity and unemployment [7].

Food security and regional impact

While President Obama referred to the USA as a whole in 2014, most food assistance is distributed at regional levels, so it is important to understand the impacts regional characteristics may have on food security. One study by Moore and Diez Roux [8] compared local food environments across regions in three states and found a number of impactful differences. Comparatively, North Carolina had the largest population and highest median income of the study, while New York was the most urban [8]. In terms of food, New York had the highest number of food stores per square mile, compared to Maryland and North Carolina, and grocery stores were the most common stores in New York and Maryland, but convenience stores were most common in North Carolina [8]. Lower-income areas were found to have more grocery stores, meat and fish markets, and liquor stores than higher-income areas but fewer supermarkets, vegetable markets, bakeries, natural food stores, and specialty stores [8]. Dharamasen et al. [7] found that food insecurity was positively correlated with race, poverty, and the number of grocery stores, but negatively correlated with expenditures at full service and fast food restaurants.

Other studies suggest that neighborhood relationships have little to no impact on food security. A study of Canadian neighborhoods was unable to link food security to neighborhood characteristics such as grocery stores, but concluded that household resource constraints and social capital contributed more to food insecurity risk [9]. Similarly, in a study done in Philadelphia, PA, researchers concluded that food access (nearness to quality food) did not impact food security, but access to aid programs and household financial constraints did [10].

Given that the economy (including food security) and health care (including non-temporary chronic illness) are leading issues in the USA, it is important to understand the overlap of household food security and health status. This study sought to quantify food security and chronic healthfulness of an extended Midwestern sample. This study focused on obtaining a sample of respondents, which was targeted to be representative in terms of sex, age, income, and state of residence. No screening of any kind related to food security was employed, which makes this data collection effort distinct from many others which focus on households with income limits or utilize screening questions (related to food security) for at least some portion of potential respondents. This study also pursued an exploration of the intersection of food security and chronic healthfulness in order to understand possible relationships between the two. This study also included a county-level poverty indicator in order to explore the relationship between community and household food security.

Methods

Measuring food security

The USDA ERS measures food security for the USA. In the “Household Food Security in the United States in 2014” report, it was reported that 14.0% of US households were food insecure or did not have “access at all times to enough food for an active, healthy life for all household members” due to limited economic resources [11].

The USDA ERS uses a survey generally referred to as the Current Populations Survey (CPS) Food Security Supplement which contains eighteen total questions: ten adult/household-specific questions and eight additional questions if children are present in the household. The questions ask whether certain behaviors involving food acquisition and diet quantity and quality occurred in the household over the last 12 months [11]. The complete eighteen questions and possible response options as presented by the CPS Food Security Supplement are included in “Appendix.” Using the ten adult/household-specific questions, food security was calculated at four levels: high food security (HFS), marginal food security (MFS), low food security (LFS), and very low food security (VLFS). Responses of “yes,” “often,” “sometimes,” “almost every month,” and “sometimes but not every month” were tallied, and each respondent was given a raw score ranging from one to ten [12]. Scores of zero were ranked as HFS, one or two as MFS, three to five as LFS, and six or higher as VLFS. When using all eighteen questions, scores of zero were ranked as HFS, one or two as MFS, three to seven as LFS, and eight or higher as VLFS [12].

An important deliberation in food security assesses the way it is measured. Burchi and De Muro [13] wrote “The way food security is theorised, measured and finally analyzed affects the typology of policies that will be adopted.” The CPS Food Security Supplement was designed to measure three food security domains: uncertainty, quality, and quantity [14, 15]. Coates et al. [14] compared 22 food security measures across 15 countries, including the CPS Food Security Supplement and several variations, and found a number of comparisons: 18 of the 22 measures include uncertainty measures, 16 of the 22 addressed the ability to eat healthy or proper diets, 21 of the 22 contained questions about running out of food and perceptions that there was not enough food for each member to eat as much as they should or want. Despite comparability, Coates et al. [14] ultimately concluded that the CPS Food Security Supplement may not adequately cover all domains important to food insecurity but offers a good foundation.

Other criteria for measuring food security have been assessed. Burchi and De Muro [13] reviewed a number of food security measures and concluded a capability approach can improve measurements by accounting for more direct and indirect drivers. However, they concede including all factors requires an ability to access large amounts of information and can be improbable [13]. A capability approach assesses food entitlements (employment status, assets, skill sets, etc.), basic capabilities (food access/ quality, health status, decision making, etc.), and the capability to be food secure (nutrition knowledge, cultural and religious beliefs, etc.) [13]. Relatedly, Diansari and Naseki [16] found that household-head education and household-head nutritional knowledge were significant variables when predicting subjective household food security; increasing either factor increased the likelihood the household would be subjectively more food secure. Headey and Ecker [17] evaluated food measures, including leading US tools, and concluded that dietary diversity is a leading food security indicator and can be used to measure trends, shocks, seasonality, and individuals, in a way that poverty and calorie availability subjective indicators (i.e., affordability and satisfaction) cannot. The CPS Food Security Supplement has a number of accepted measurement criteria [14, 17], but could be improved with more information.

Survey instrument and data collection

A survey instrument was designed to collect data for this analysis and was hosted at Purdue University using Qualtrics. Lightspeed GMI facilitated obtaining a sample of Midwest residents through a large proprietary opt-in database in February 2016. The respondent’s sex, age, annual pretax household income, and state of residence were targeted to be representative of the population of the Midwest region based on the U.S. Census Bureau [18]. The survey attracted 7277 total link clicks, 7 people did not start the survey, 29 were under the age of 18, 578 were pushed out because of the state of residence quota, 5377 were pushed out because of the income quota, and 21 gave questionable or extreme response upon evaluation of the data. The final sample of completed surveys used in this analysis was 1265.

This study defines the Midwest according to the U.S. Census Bureau region definition and represents the states of Illinois, Indiana, Iowa, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin [19], but also includes Midwestern neighbors Kentucky and Tennessee. Looking at the 2012–2014 state averages presented by Coleman-Jensen et al. [11], the Midwest contained a wide range of food insecurity proportions, with North Dakota having the lowest Midwestern and national rate of 8.4%, and Ohio with the highest of the Midwest at 16.9%. Kentucky and Tennessee had rates of 17.5 and 16.3% food insecurity, respectively [11]. A comparison of food insecurity rates for this study’s sample and those reported by Coleman-Jensen et al. [11] are provided in the results.

Demographic questions included in the survey asked about sex, age, annual pretax household income, state of residence, and household composition. For annual pretax household income, seven income categories were included and were condensed into three categories: low income: less than $35,000, mid-income: $35,000–$75,000, and high income: $75,000 or more. County of residence was collected in addition to respondents’ state of residence. Each county was paired with a U.S. Census Bureau Poverty estimate [20] in order to evaluate community conditions, in relationship with household specifics. The estimate represents the percent of the county population of all ages living at or below the national poverty line. For this study, the proportions were grouped into three categories 0–10, 11–20, and 21% or greater. Like the food insecurity proportions presented in Coleman-Jensen et al. [11], the Midwest has a wide range of county-level poverty. For this sample, North Dakota had the lowest county average poverty of 10.8% and Ohio with the highest of 16%, and Kentucky had the highest rates of the sample with a county average of 20.6% living in poverty.

The food security questions explored were based on the CPS Food Security Supplement [11]. All respondents were asked the ten adult/household-specific questions. The results were calculated using the methods outlined in the “U.S. Household Food Security Survey Module: Three-stage Design, with Screeners” report [21]. The report suggests a screening question and income-based screening options. These options were left out of this study in order to obtain a broader understanding of the food security status of the complete sample of respondents.

For comparative purposes, two subsamples, households with and without children, were generated. All respondents were asked the ten household focused food security questions, and the sample of respondents who indicated that there were no children living in the household (n = 749) were asked no further questions. Respondents who indicated having children (persons younger than 18 years old) (n = 360) formed a second subsample and were asked the full eighteen-question survey, and the four levels were calculated with the score adjusted to accommodate the added questions. No supplementary calculations were done for the remaining respondents who did not indicate clearly if children were present in the household (n = 150).

Respondents were asked questions about the prevalence of health conditions in their households. To assess health conditions, respondents were asked “Please indicate if you or someone in your household have any of the following conditions” which included a list of responses: diabetes, Crohn’s disease, celiac disease, eating disorder, depression/ anxiety, high blood pressure, and high cholesterol.

Data analysis

Summary statistics were generated for each question, and cross-tabulations were performed in SPSS statistical software [22]. An ordered logit model was estimated using Stata/SE 14.1 [23] in order to identify determinates that contribute to the likelihood that a respondent will fall into increasing levels of food insecurity severity.

Ordered logits have been used to assess the probability of ranked outcomes. Migliore et al. [24] used an ordered logit to evaluate the likelihood of increased organic food purchases using quality conventions and income as independent variables. Diansari and Nanseki [1] used an ordered logit to predict household subjective food security status in Indonesia using insecure, somewhat insecure, somewhat secure, secure, and highly secure as the ranked dependent variable. Peterson et al. [25] employed an ordered logit to understand consumer factors (credence attributions, time, conveniences) influencing the likelihood a consumer will choose a specific local food retailer, among US and French consumers.

For each respondent, the food security level was converted into a numeric value in order to generate a discrete dependent variable: HFS equaled zero, MFS one, LFS two, and VLFS three. The independent variables were primarily discrete binary variables where one equaled the variable descriptor. The variables were: male, age 18–24, age 25–44, age 45–64, low income less than $35,000, mid-income: $35,000–$74,999, diabetes, Crohn’s disease, celiac disease, eating disorder, depression/anxiety, high blood pressure, and high cholesterol. Percent of county population in poverty was the only continuous variable with the potential to range from 0 to 100%.

Ordered logits estimate the likelihood that an outcome will fall between or beyond estimated thresholds and the thresholds were calculated using the ranked dependent variable [26]. For this study, y represents the dependent variable and can take on the values: y = 0, 1, 2, or 3.

Rank would be established using k to represent thresholds of j,
$$\begin{aligned} & {\text{If}}\,\,k_{j - 1} < y^{* } < k_{j } \quad {\text{then}}\,y = j, \quad {\text{for}}\,j = 1, \ldots ,2, \,{\text{and}} \\ & {\text{If}}\,\,k_{2} < y^{* } \quad {\text{then}} \,y = 3, \\ \end{aligned}$$
where y* is a latent variable [26] and will be estimated by the model. The food security rank for each respondent is represented by y*. For each survey respondent i, food security can be explained by variables Xi,
$$y^{* } = \beta X_{i} + u_{i} .$$
The probability of each rank, j, can then be estimated and depends on the regression outcome falling between kj and kj−1 [26]
$$Pr\left( {y_{i} = j} \right) = Pr\left(k_{j - 1} < \beta X_{i} + u_{i} < k_{i}\right).$$

Results

A sample targeted to be representative of the Midwest population was collected in February of 2016 and consisted of 1265 completed surveys. Summary statistics on demographics reported are provided in Table 1. Forty-nine percent of the sample was male, and 28% of households reported having children. Income was collected by providing seven principal categories, but for conciseness the income categories were condensed into three groups: low income: less than $35,000 (32%), mid-income: $35,000–$75,000 (34%), and high income: more than $75,000 (34%).
Table 1

Sample demographics (n = 1265).

Source: Population percentages obtained from: U.S. Census Bureau; American Community Survey, 2014 American Community Survey 1-Year Estimates, Table S1901; generated by S. R. Dominick; using American FactFinder; http://factfinder2.census.gov; 21 Sept 2015

Variable description

Survey (% of respondents)

U.S. Census Bureau, 2014 American Community Survey 1-year estimates (%)

Male

48

49

Age

18–24

08

13

25–44

33

31

45–64

38

36

65 and older

21

20

Income

Less than $25,000

21

24

$25,000–$34,999

11

11

$35,000–$49,999

14

14

$50,000–$74,999

20

19

$75,000–$99,999

13

12

$100,000–$149,999

13

12

$150,000 or more

09

08

Household composition

No children in household

59

 

Children in household

28

 

Unstated or ambiguous

12

 

Poverty level % of county population

Low: less than 10%

22

 

Mid: 11–20%

64

 

High: more than 21%

13

 

You or someone in your household have any of the following conditions

Diabetes

24

 

Crohn’s disease

08

 

Celiac disease

07

 

Eating disorder

08

 

Depression/anxiety

25

 

High blood pressure

44

 

High cholesterol

41

 

U.S. Census Bureau Poverty estimates for each county [20] were evaluated and condensed into three categories. Twenty-two percent of respondents lived in a county where less than 10% of the population lived at or below the poverty line (low poverty), 64% lived where 11–20% of the population lived at or below the poverty line (mid-poverty), and 13% lived where those living at or below poverty make up 21% or more of the population (high poverty). Survey participants were asked about a number of chronic health conditions. Forty-four percent of respondents indicated that they or someone in their household had high blood pressure. Personal or household member high cholesterol was selected by 41%, depression/anxiety by 25%, diabetes by 24%, Crohn’s disease and eating disorders were each selected by 8% of the sample, and celiac disease was selected by 7%.

Table 2 summarizes the percent of respondents from each state who were calculated to be in food insecure households and compares that proportion with the estimations reported by the USDA [11]. For most states, the percent of respondents who were found to be in food insecure households were higher than those estimated by the USDA. Comparatively, the highest state food insecurity rate estimated by both this study and the USDA was for Kentucky and the lowest for North Dakota. The results for this sample could be higher because the respondents were not screened by income or using the optional screening question [12]. Table 2 also summarizes the percent of the population of each state living at or below the poverty line, for which the highest rate is also in Kentucky and the lowest are North Dakota and Minnesota.
Table 2

State-by-state food insecurity comparison with 2012–2014 USDA estimated averages

 

% of sample respondents per state (n = 11,265)

USDA ERS 2012–2014 state averages (%)

% of county population living at or below poverty (sample averages)

Illinois

35

11.7

14

Indiana

24

14.6

15

North Dakota

00

8.4

11

South Dakota

21

11.9

15

Ohio

24

16.9

16

Kansas

28

15.9

13

Nebraska

16

13.9

12

Iowa

06

11.4

12

Missouri

31

16.8

14

Michigan

27

14.7

16

Minnesota

14

10.4

11

Wisconsin

14

16.3

13

Kentucky

39

17.5

21

Tennessee

27

11.4

18

Mean

22

14

 

Source of Comparison USDA ERS 2012–2014 Average: Coleman-Jensen et al. [11]. Household Food Security in the United States in 2014, ERR-194, U.S. Department of Agriculture, Economic Research Service

Analyzing food security status across household characteristics

A food security status was calculated for the survey sample (n = 1265) as well as for two subsamples (households with reported children n = 360 and households without children n = 749). A summary of the proportions can be found in Fig. 1. Sixty-five percent of the total sample of HFS, 25% were food insecure, with 10% of LFS and 15% of VLFS. For households with children, 44% were food insecure compared to 18% of households without children. Households with children had the highest proportion of respondents in each food insecure group, 15% were of LFS, and 28% were of VLFS. Households without children had 8% of LFS and 10% of VLFS.
Fig. 1
Fig. 1

USDA food security score (% of respondents n = 1265). High food security and food insecure do not sum to 100 (due to the marginal food security category). The difference in “households with Children” and “households without children” can be accounted for by a group of respondents for whom specific household make up could not be determined, and these are known as ambiguous households (n = 156)

The total sample and the two household composition subsamples food security statuses, not including MFS for concision, were cross-tabulated with four demographic categories: sex, age, income level, and percent population living at or below the poverty line for the respondent’s county of residence. A full summary can be found in Table 3.1 Z tests were performed to test the statistical difference in proportions across demographic categories. For additional understanding, each of the ten questions and subsequent responses were cross-tabulated with the same demographics and can be found in Table 4.
Table 3

USDA food security status and demographics cross-tabulation

Food security level

Sex

Age

Income levels

Poverty level % of county pop.

Male (A)

Female (B)

18–24 (C)

25–44 (D)

45–64 (E)

65 + (F)

Low (G)

Middle (H)

High (I)

Low (J)

Mid (K)

High (L)

All Adults

Count

607

658

96

415

485

269

407

425

433

280

815

170

High food security

61.8B

67.9A

51.0EF

48.7EF

70.9CDF

84.4CDE

47.2HI

68.7GI

78.1GH

72.9KL

63.3J

60.0J

Food insecure

28.5B

21.9A

35.4EF

40.2EF

18.1CDF

10.4CDE

34.4HI

22.1G

19.2G

17.5KL

26.0J

31.7J

Low food security

9.7

9.6

16.7EF

13.0EF

7.4CD

5.9CD

14.5I

10.4I

4.4GH

6.8

10.4

10.6

Very low food security

18.8B

12.3A

18.8EF

27.2EF

10.7CDF

4.5CDE

19.9HI

11.8G

14.8G

10.7KL

15.6JL

22.4JK

Households with children

Count

190

170

28

250

75

7

52

132

176

77

239

54

High food security

38.4B

53.5A

32.1EF

42.8EF

56.0CD

85.7CD

17.3HI

44.7G

54.5G

53.2

43.7

42.6

Food insecure

48.9B

37.6A

57.1EF

46.4E

32.0CD

14.3C

59.6HI

40.2G

41.5G

35.1

45.4

48.1

Low food security

12.6

18.2

32.1D

13.2C

16.0

14.3

34.6HI

18.2GI

7.4GH

19.5

14.0

14.8

Very low food security

36.3B

19.4A

25.0

33.2E

16.0D

0.0

25.0

22.0I

34.1H

15.6KL

31.4J

33.3J

Households without children

Count

343

406

46

140

341

222

297

243

209

166

486

97

High food security

72.3

72.4

56.5EF

58.6EF

71.8CDF

85.1CDE

51.2HI

80.2GI

93.3GH

79.5KL

70.8J

68.0J

Food insecure

19.0

17.5

30.4EF

30.7EF

16.7CDF

9.9CDE

31.0HI

14.4GI

4.3GH

10.2KL

18.9JL

27.8JK

Low food security

7.6

8.4

15.2EF

15.0EF

5.9CD

5.4CD

13.5HI

7.0GI

1.4GH

2.4KL

9.3J

11.3J

Very low food security

11.4

9.1

15.2F

15.7F

10.9F

4.5CDE

17.5HI

7.4GI

2.9GH

7.8L

9.7L

16.5JK

For each demographic category (i.e., sex, age), each value of one column marked with a superscript of another column denotes significant difference at the .05 level. No superscript denotes no difference. No calculations were done across demographic categories. High food security and food insecure do not sum to 100. The difference is for a category called marginal food security. The difference in “households with Children” and “households without children” can be accounted for by a group of respondents for whom specific household make up could not be determined; these are known as ambiguous households (n = 156)

Table 4

USDA 10 demographic, income, and poverty comparisons

 

% of resp.

Sex

Age

Income levels

Poverty level % of county pop

Male (A)

Female (B)

18–24 (C)

25–44 (D)

45–64 (E)

65 + (F)

Low (G)

Middle (H)

High (I)

Low (J)

Mid (K)

High (L)

Count

 

607

658

96

415

485

269

407

425

433

280

815

170

We worried whether our food would run out before we got money to buy more

Often true

08

9.6

7.3

9.4F

16.1EF

4.7D

2.6CD

13.0HI

4.5GI

7.9GH

3.9KL

8.7JL

14.1JK

Sometimes true

17

20.8B

14.4A

22.9DF

24.8EF

15.7DF

7.4CDE

24.8HI

16.9GI

11.1GH

14.3

18.3

18.8

Never true

72

66.9B

76.3A

62.5EF

55.9EF

77.9CDF

88.5CDE

57.5HI

76.9G

80.1G

80.0KL

70.4J

64.7J

The food that (I/we) bought just didn’t last, and (I/we) didn’t have money to get more.

Often true

08

10.9B

6.2A

7.3DF

18.3CEF

3.5D

2.6CD

8.8H

5.2GI

11.3H

4.3KL

8.5JL

15.3JK

Sometimes true

17

18.5

15.5

31.3EF

22.2EF

15.1CDF

7.1CDE

27.3HI

15.8GI

8.3GH

12.9

17.8

19.4

Never true

72

67.5B

76.9A

58.3EF

55.7EF

80.2CDF

89.2CDE

60.4HI

76.9G

79.2G

81.4KL

71.0J

64.1J

I/we couldn’t afford to eat balanced meals.

Often true

08

8.6

7.9

9.4F

13.5EF

7.0FD

1.9CDE

13.5HI

4.9G

6.5G

4.6 K

9.4J

8.2

Sometimes true

19

21.1

17.8

26.0EF

28.9EF

15.3CDF

9.7CDE

28.3HI

18.8GI

11.5GH

15.4L

19.4

25.9J

Never true

70

67.5B

72.8A

59.4EF

54.2EF

76.5CDF

87.7CDE

55.0HI

74.1GI

80.8GH

78.2KL

68.7J

64.7J

Did you (or other adults in your household) ever cut the size of your meals or skip meals because there wasn’t enough money for food?

Yes

18

20.3

16.1

20.8F

29.9EF

13.4DF

7.4CDE

26.0HI

14.1G

14.5G

12.9KL

18.7J

24.1J

No

79

76.4B

81.3A

71.9EF

65.8EF

84.9CDF

91.1CDE

68.6HI

83.5G

84.3G

85.0KL

77.8J

74.7J

Of those who said yes: How often did this happen?

Almost every month

47

60.2B

32.1A

40.0

58.9E

27.7D

45.0

34.0I

40.0I

76.2GH

47.2

44.7

56.1

Some months

38

30.1B

48.1A

45.0

29.8E

52.3D

40.0

45.3I

45.0I

20.6GH

33.3

40.8

34.1

Only 1 or 2 months

13

7.3B

18.9A

5.0

11.3

16.9

15.0

17.9I

13.3I

3.2GH

11.1

13.8

9.8

Did you eat less than you felt you should because there wasn’t enough money for food?

Yes

18

19.6

17.2

26.0EF

25.8EF

15.1CD

10.0CD

27.8HI

15.8G

12.0G

14.3L

18.9

22.4J

No

79

77.3

80.1

67.7EF

69.4EF

82.9CDF

89.6CDE

67.3HI

81.4GI

86.8GH

83.2L

77.9

75.3J

You were hungry but didn’t eat because there wasn’t enough money for food?

Yes

14

16.3B

11.2A

18.8EF

22.7EF

10.3CDF

4.1CDE

18.9HI

12.0G

10.4G

9.6KL

14.2J

17.6J

No

83

79.2B

85.7A

75.0EF

71.3EF

87.4CDF

94.1CDE

74.9HI

84.5G

88.0G

87.1KL

81.3J

81.2J

Did you lose weight because there wasn’t enough money for food?

Yes

10

14.0B

6.8A

10.4F

16.9EF

8.0DF

4.1CDE

12.0

8.5

10.4

7.9L

10.2

14.7J

No

86

81.4B

89.8A

79.2EF

77.1EF

89.5CDF

94.8CDE

80.8HI

88.0G

88.2G

88.6L

85.8

81.2J

Did you or other adults in your household ever not eat for a whole day because there wasn’t enough money for food?

Yes

09

11.9B

5.8A

8.3DF

17.8CEF

4.7DF

1.9CDE

8.4

7.8

9.9

4.6KL

9.2J

12.9J

No

89

84.8B

92.1A

87.5DEF

77.6CEF

93.6CDF

97.0CDE

87.2

89.9

88.7

93.2KL

88.0J

84.1J

Of those who said yes: How often did this happen?

Almost every month

43

43.1

42.1

12.5D

50.0CE

26.1D

60.0

38.2

39.4

48.8

38.5

42.7

45.5

Some months

47

44.4

52.6

87.5D

43.2C

47.8

40.0

47.1

48.5

46.5

46.2

48.0

45.5

Only 1 or 2 months

05

8.3

0.0

0.0

2.7E

17.4D

0.0

8.8

6.1

2.3

0.0

6.7

4.5

For each demographic category (i.e., sex, age), each value of one column marked with a superscript of another column denotes significant difference at the .05 level. Responses for the category “don’t know/ prefer not to answer can be found by subtracting the sum of responses shows per question from 100%

Females received HFS scores of zero more frequently than males. When looking at the ten adult/household-specific questions, males answered “often true,” “almost every month,” and “yes” more frequently when compared to females.2 Sixty-seven percent of females had a calculated score of zero compared with 61.8% of males. Of the total sample, 18.8% of males and 12.3% of females reported being of VLFS. In terms of specific severity, males more frequently selected “yes” in response to the questions “You were hungry but didn’t eat because there wasn’t enough money for food,” “Did you lose weight because there wasn’t enough money for food,” and “Did you or other adults in your household ever not eat for a whole day because there wasn’t enough money for food?” The trends were similar for households with children. While there is less security overall for each sex, a greater portion of females (53.5%) were considered to be of high food security compared with males (38.4%).

For the total sample, younger respondents had higher rates of food insecurity than older respondents. Thirty-five percent of 18–24-year-olds and 40.2% of 25–44 had a calculated score of three or higher and were considered food insecure, compared with 18.1% of 45–64-year-olds and 10.4% of those 65 years old or older. For the statement “The food that (I/we) bought just didn’t last, and (I/we) didn’t have money to get more,” 18.3% of 25–44-year-olds selected “Often true,” and was over double the frequency of 18–24-year-olds (7.3%), 45–64-year-olds (3.5%), and those 65 years old or older (2.6%). The youngest age groups also had the highest proportions of VLFS, 18.8% of 18–24-year-olds and 27.2% of 25–44-year-olds, compared with 4.5% of those 65 years old or older. Distinctively, for the questions “Did you (or other adults in your household) ever cut the size of your meals or skip meals because there wasn’t enough money for food,” “Did you eat less than you felt you should because there wasn’t enough money for food?” “You were hungry but didn’t eat because there wasn’t enough money for food,” the younger two categories selected “yes” at higher rates compared to the older two categories.

The subsample of households with children showed similar trends with the youngest age groups more frequently in food insecure categories. Respondents considered of HFS made up 32.1% of 18–24-year-olds and 42.8% of 25–44-year-olds compared with 56.0% of 45–64-year-olds and 85.7% of those 65 years old or older. Respondents 25–44 had the largest portion of VLFS individuals with 33.2%, and this was not different from the 25% of 18–24-year-olds but was different and higher than those 45–64 years old.

Less of a gap between high food security and food insecure respondents was found across ages for the subsample of households without children. Fifty-six percent of 18–24-year-olds and 58.6% of 25–44-year-olds were considered of HFS, compared with 71.8% of 45–64-year-olds and 85.1% of those 65 years old or older. Of those who reported being 18–24 and being 25–44 years old, 30.4 and 30.7% were considered food insecure, compared with 16.7% of respondents 45–64 years old and 9.9% of 65-year-olds or older. The youngest age groups also had the highest proportions for LFS and VLFS.

Respondents in the low- and middle-income levels, largely, selected “often true,” “almost every month,” and “yes” with higher frequency than the remaining income groups. The low-income group more frequently selected “often true” and “sometimes true” for the statement “We worried whether our food would run out before we got money to buy more,” but, the high-income group was the next most frequent selector of “often true” and the lowest selector of “sometimes true.” For the statement “The food that (I/we) bought just didn’t last, and (I/we) didn’t have money to get more.” Both the low-income group (8.8%) and the high-income group (11.3%) selected “often true” more frequently compared to the mid-income (5.2%) and were not statistically different from each other. The results for the question “Did you (or other adults in your household) ever cut the size of your meals or skip meals because there wasn’t enough money for food?” and the sub question of “how often did this happen?” are particularly interesting. While the low-income group more frequently selected “yes,” the high-income group more frequently selected “almost every month.”

Looking at the total sample, the low-income group had the lowest proportion of HFS respondents and the highest portions of LFS and VLFS respondents. Thirty-four percent of the respondents in the low-income group were considered food insecure; specifically, 14.5% were of LFS and 19.9% of VLFS. For the mid-income group, 22.1% were food insecure, which was not different from the 19.2% of high-income respondents who were food insecure. Also 10.4% of mid-income respondents were of LFS and 11.8% were of VLFS. The high-income group had the largest proportion of respondents in the HFS group with 78.1%.

The lowest-income group still reported the largest proportions of food insecurity, but the proportions for mid- and high incomes were also larger when looking specifically at the subsample of households with children. Fifty-nine percent of the low-income group, 40.2% of the mid-income group, and 41.5% of the high-income group were considered food insecure. Interestingly, one of the higher proportions of VLFS for households with children was reported in the high-income group. Thirty-four percent of the high-income group had a calculated VLFS status, and was not different from the 25.0% reported in the low-income group, but is different from the proportion of mid-income respondents with 22.0%. When looking at households without children, the low-income group reported the highest food insecure proportion. HFS respondents made up 51.2% of the low-income group and 31.0% were food insecure, compared with 93.3 and 4.3% of the high-income group, respectively. The low-income group also reported the highest LFS and VLFS proportions with 13.5 and 17.5%.

Respondents living in a county with a higher percent of people living at or below poverty have higher rates of food insecurity. Seventy-two percent of the total sample living in areas of low poverty were considered of HFS, compared to 63.3% of respondents living at mid-poverty and 60.0% of respondents living at high poverty. Respondents living in a high-poverty county were more frequently of VLFS (22.4%) compared with those living in a low-poverty county (10.7%) and a mid-poverty county (15.6%). Those living in a high-poverty county also selected “yes” more frequently to the questions “Did you eat less than you felt you should because there wasn’t enough money for food?” “You were hungry but didn’t eat because there wasn’t enough money for food?” “Did you lose weight because there wasn’t enough money for food?”

For the subsample of households with children, 31.4% of respondents living in a mid-poverty county and 33.3% of respondents living in a high-poverty county were of VLFS, almost double the rate of respondents living in a low-poverty county (15.6%). Households without children living in a low-poverty county were more frequently of HFS, 79.5% compared with 70.8% of respondents living in a mid-poverty county, and 68.0% living in a high-poverty county. Respondents living in a high-poverty county were also more frequently food insecure with 27.8% of the sample and were more frequently of VLFS (16.5%).

Food security and chronic health indicators

If food aid programs are intended to support temporary conditions, it is important to consider the impact of chronic health conditions on food security status. Respondents were asked to indicate whether they or someone in their household had any of seven common conditions, which may impact an afflicted person’s relationship with food. The conditions were: diabetes, Crohn’s disease, celiac disease, eating disorder, depression/ anxiety, high blood pressure, and high cholesterol.

Figure 2 displays a preliminary analysis of the hypothesis that a relationship exists between chronic health and food security, even among the high-income groups. Each income level was proportioned into two groups: those for whom someone in the household had at least one of the health conditions and those for whom no one is afflicted. Each of those six groups was then proportioned into three levels of food security: HFS, MFS, and food insecure.
Fig. 2
Fig. 2

Proportionality of food security among respondents with and without chronic health conditions (% of respondents n = 1265)

For the low-income group, 132 respondents did not report any instances of the seven chronic health conditions and 275 reported at least one. Of those who did not report any instances, 54% were of HFS, 11% were of MFS, and 36% were food insecure. Of the 275 respondents that reported a chronic health condition, 44% were of HFS, 22% were of MFS, and 34% were food insecure. In the mid-income group, 163 respondents reported no chronic health conditions and 262 reported at least one. Of those who did report a health condition, 68% were considered of HFS, 10% were of MFS, and 23% were considered food insecure. Of those who did not report a condition, 71% were of HFS, 9% were of MFS, and 21% were considered food insecure. Among the high-income group, 139 respondents did not report a health condition and 294 indicated at least one. Of those who did not report a condition, 91% were considered of HFS, 0% were of MFS, and 9% were considered food insecure. Of those who did report a condition, 72% were of HFS, 4% were of MFS, and 24% were considered food insecure. Since this population was of particular interest, a comparative Fishers exact test was performed and revealed that the proportions of respondents who were high income, had a household health condition, and were food insecure were statistically different and higher compared to those who were of high income, had no household health conditions, and were food insecure. In short, the intersection between food insecurity and health conditions consistently emerges as important and necessary to consider if aiming to improve household’s well-being, and this consideration exists even among high-income groups.

To understand the relationship between food security and health, cross-tabulations and Z test were performed to determine statistical difference between the proportions of respondents (Table 5). Generally, higher rates of food insecurity existed for respondents who indicated that at least one health condition was present in the household.
Table 5

USDA food security status and chronic condition cross-tabulation

Food security level

Diabetes

Crohn’s disease

Celiac disease

Eating disorder

Depression/ anxiety

High blood pressure

High cholesterol

No (A)

Yes (B)

No (A)

Yes (B)

No (A)

Yes (B)

No (A)

Yes (B)

No (A)

Yes (B)

No (A)

Yes (B)

No (A)

Yes (B)

All adults

Count

958

307

1167

98

1181

84

1165

100

949

316

709

556

752

513

High food security

69.8B

49.8A

68.1B

27.6A

67.8B

25.0A

68.3B

26.0A

72.4B

42.7A

65.6

64.2

65.6

64.1

Food insecure

20.9B

38.1A

21.9B

62.2A

22.2B

65.5A

21.7B

64.0A

19.6B

41.5A

25.2

24.8

25.5

24.4

Low food security

9.4

10.4

9.5

11.2

9.6

10.7

9.4

13.0

8.6B

12.7A

11.4B

7.4A

11.8B

6.4A

Very low food security

11.5B

27.7A

12.4B

51.0A

12.6B

54.8A

12.4B

51.0A

11.0B

28.8A

13.8

17.4

13.7B

17.9A

Households with Children

Count

249

111

293

67

297

63

300

60

244

116

236

124

246

114

High food security

55.8B

22.5A

52.2B

16.4A

52.2B

14.3A

52.0B

13.3A

54.1B

27.6A

53.0B

31.5A

52.0B

31.6A

Food insecure

33.3B

66.7A

35.2B

80.6A

35.4B

82.5A

35.7B

83.3A

35.2B

61.2A

36.0B

58.1A

38.2B

55.3A

Low food security

15.7

14.4

16.0

11.9

16.5

9.5

16.3

10.0

14.8

16.4

16.5

12.9

17.1

11.4

Very low food security

17.7B

52.3A

19.1B

68.7A

18.9B

73.0A

19.3B

73.3A

20.5B

44.8A

19.5B

45.2A

21.1B

43.9A

Households without children

Count

592

157

726

23

735

14

594

19

593

156

396

353

426

323

High food security

73.8

66.9

72.9

56.5

72.4

71.4

73.4B

48.4A

77.6B

52.6A

70.2

74.8

71.8

73.1

Food insecure

16.7B

23.6A

17.8

30.4

18.2

14.3

17.3B

38.7A

14.3B

32.7A

20.5

15.6

19.7

16.1

Low food security

7.8

8.9

8.0

8.7

8.0

7.1

7.7

16.1

6.6B

13.5A

9.3

6.5

10.6B

4.6A

Very low food security

9.0B

14.6A

9.8

21.7

10.2

7.1

9.6B

22.6A

7.8B

19.2A

11.1

9.1

9.2

11.5

For each demographic category (i.e., sex, age), each value of one column marked with a superscript of another column denotes significant difference at the .05 level. No superscript denotes no difference. No calculations were done across demographic categories. High food security and food insecure do not sum to 100. The difference is for a category called marginal food security. The difference in “households with Children” and “households without children” can be accounted for by a group of respondents for whom specific household make up could not be determined, and these are known as ambiguous households (n = 156)

Respondents who indicated having diabetes or that someone in the household had diabetes had larger proportions of food insecurity overall and larger proportions of VLFS. These relationships were stronger for the subsample of households with children. Of households with children, 66.7% of respondents who reported household diabetes were food insecure, compared with 33.3% of respondents without household diabetes. Over half the respondents with household diabetes were of VLFS, 52.3% compared with only 17.7% of respondents with no instance of diabetes. The subsample of households without children has a similar relationship. Twenty-three percent of respondents with household diabetes were food insecure, compared with 16.7% of respondents without an instance of diabetes. Also 14.6% of respondents with household diabetes were of VLFS, compared with 9.0% without household diabetes.

Respondents with household Crohn’s disease were more frequently food insecure (62.2%) than those without household Crohn’s disease (21.9%). Over half the respondents with Crohn’s disease were of VLFS (51.0% compared with 12.4%). This relationship is also true for the subsample of respondents with children. Eighty percent of respondents with children and with household Crohn’s disease were also food insecure, and 68.7% were of VLFS. Households with children but without instances of Crohn’s disease were HFS, at 52.2%. For households with celiac disease, 65.5% were food insecure and 54.8% were of VLFS. Comparatively, 67.8% of respondents without instances of celiac disease were HFS. For the subsample of households with children, 82.5% were food insecure and 73% were of VLFS when celiac disease was also present in the household.

Of respondents who indicated they or someone in their household experienced an eating disorder, 26.0% were HFS and 64.0% were food insecure, in contrast to respondents who did not report household eating disorders (68.3% were HFS). Over half the respondents who indicated household eating disorders were also of VLFS (51.0%). Similarly, of households with children, 52.0% of respondents who did not report an eating disorder were HFS and 83.3% of households which reported an eating disorder were food insecure, with 73.3% of VLFS. In households without children, the trend is similar, though less extreme. Only 38.7% of households without children who reported household eating disorders were food insecure, and 22.6% were of VLFS. Comparatively, 17.3% of respondents who did not report an eating disorder were food insecure and 9.6% were of VLFS.

Of respondents who reported household depression/anxiety, 42.7% were considered HFS, while 72.4% of respondents who did not report household depression/anxiety were considered HFS. Respondents who reported household depression/anxiety were also more frequently of LFS (12.7%) and of VLFS (28.8%) compared with respondents who did not report household depression/anxiety (8.6 and 11.0%, separately). Similarly, of the subsample of households with children, 44.8% of those who reported household depression/anxiety were of VLFS and 61.2% were food insecure in total. More frequently, households without reported depression/anxiety were food secure (54.1% compared with 27.6%). Households without children who reported household depression/anxiety were more frequently food insecure. Thirty-two percent of households without children who reported household depression/anxiety were food insecure compared with 14.3% of households without children and without depression/anxiety. Also, more respondents without children were frequently of VLFS, 19.2% compared with 7.8%.

For the total sample, food security for respondents with household high blood pressure was not statistically different from respondents without household instance of high blood pressure except in the category LFS. Eleven percent of respondents without household high blood pressure were more frequently of LFS, compared with 7.4% of respondents with household high blood pressure. This result is counterintuitive to the results presented for the previous conditions. Households with children showed results similar to all other reported conditions. Of households with children, 53.0% of respondents who did not indicate instance of high blood pressure were HFS, compared with 31.5% of respondents who did indicate. Fifty-eight percent of respondents with household high blood pressure were food insecure and 45.2% were of VLFS.

Respondents without instances of high cholesterol were more frequently considered of LFS (11.8%) compared with respondents who reported household high cholesterol (6.4%). Conversely, respondents with household high cholesterol were more frequently considered of VLFS. Households with children were more frequently food insecure and of VLFS if they also reported that they or someone in their household had high cholesterol. Respondents without instance of household high cholesterol were more frequently HFS. Respondents in households without children and with high cholesterol in the household were statistically different from, and less frequently of LFS, than households without children and without household instance of high cholesterol.

Identifying determinants of food insecurity status with ordered logit model

An ordered logit was run in order to understand the significance of household chronic health status as a contributing factor to short-term (12-month) food security status, and the results can be found in Table 6. The model run was found to be significant overall, and the likelihood ratio χ2 results are presented.
Table 6

Estimated level of food insecurity ordered logit (n = 1265)

Variables

Coefficient (SE)

Marginal effects

High food security

Marginal food security

Low food security

Very low food security

dy/dx (SE)

dy/dx (SE)

dy/dx (SE)

dy/dx (SE)

Male

.2498* (.1315)

− .0522 (.0275)

.0149 (.0079)

.0159 (.0084)

.0214 (.0113)

Age 1824

1.1435*** (.2971)

− .2393 (.0616).

.0684 (.0190)

.0728 (.0196)

.0980 (.0256)

Age 2544

1.7554*** (.2419)

− .3673 (.0494)

.1050 (.0177)

.1117 (.0175)

.1504 (.0213)

Age 4564

.9182*** (.2150)

− .1921 (.0444)

.0549 (.0139)

.0584 (.0143)

.0787 (.0184)

Low income less than $35K

2.1023*** (.1903)

− .4399 (.0377)

.1258 (.0168)

.1338 (.0156)

.1802 (.0175)

Middle income: $35K$74,999

.6972*** (.1809)

− .1459 (.0373)

.0417 (.0116)

.0444 (.0118)

.0597 (.0155)

Household with children

.9298*** (.1614)

− .1945 (.0335)

.0556 (.0110)

.0592 (.0112)

.0797 (.0142)

Poverty level % of county population

.0225* (.0124)

− .0047 (.0026)

.0013 (.0007)

.0014 (.0008)

.0019 (.0010)

Diabetes

.5086*** (.1736)

− .1064 (.0362)

.0304 (.0108)

.0323 (.0113)

.0436 (.0150)

Crohn’s disease

.3183 (.3896)

− .0666 (.0815)

.0190 (.0233)

.0202 (.0248)

.0272 (.0334)

Celiac disease

.1444 (.4642)

− .0302 (.0971)

.0086 (.0277)

.0091 (.0295)

.0123 (.0398)

Eating disorder

.9906*** (.3497)

− .2072 (.0732)

.0592 (.0217)

.0630 (.0229)

.0849 (.0302)

Depression/anxiety

.5097*** (.1543)

− .1066 (.0323)

.0305 (.0097)

.0324 (.0101)

.0436 (.0134)

High blood pressure

− .0188 (.1654)

.0039 (.0346)

− .0011 (.0099)

− .0012 (.0105)

− .0016 (.0141)

High cholesterol

.1336 (.1715)

− .0279 (.0358)

.0079 (.0102)

.0085 (.0109)

.0114 (.0147)

Cut one

3.9113 (.3257)

    

Cut two

4.5523 (.3329)

    

Cut three

5.3128 (.3413)

    

Pseudo-R2

.1588

    

Prob > χ2

.0000

    

Log likelihood

−1089.1399

    

For STATA, the ordered logit thresholds are reported as “cuts.” Margins predicted at sample means

p values: * p < .10, ** p < .05; *** p < .01

All demographics included were significant determinates. Being male increased the likelihood the household would have increased food insecurity. At the margin, being male decreased the likelihood of a household receiving a score of zero (or being high food security) and increased the likelihood of receiving a score greater than zero. Being in either of the included age categories increased the likelihood that a respondent would be food insecure, and being 25–44 had the greatest marginal contributions. The lower-income categories were also significant variables for predicting food insecurity and being in the low-income group had the higher marginal contributions. The presence of a child in the household increased the likelihood that a respondent would be food insecure.

The poverty-level score was the only continuous variable, and the logit suggests that as the poverty level of the county increases, a household living in that county was more likely to be food insecure. Of all seven chronic illnesses, diabetes, eating disorders, and depression/anxiety were statistically significant. Having either of these conditions present in the household increases the likelihood that a respondent will be food insecure. At the margin, having an eating disorder had the highest contribution. All of the logit results supported the findings of the cross-tabulations.

Discussion

The results of the cross-tabulations for income were consistent with findings from the USDA in that lower incomes generally experience reduced food security. However, there were areas where closer examination of each question suggested reduced security could be experienced across broader income ranges. Recall, there were comparable results and statistical sameness between the high- and low-income groups for the statements “We worried whether our food would run out before we got money to buy more,” “The food that (I/we) bought just didn’t last, and (I/we) didn’t have money to get more” and for the question “Did you (or other adults in your household) ever cut the size of your meals or skip meals because there wasn’t enough money for food?” and the sub question of “how often did this happen?”

In order to consider the relationship between high income and food security further, the household composition of high-income respondents was investigated. For high-income respondents whose household compositions were known (n = 385), respondents were divided into two groups, those in a household with four or more people (n = 140) and those with a household composition of less than four (n = 245). A Fisher’s exact test was performed, and high-income respondents with a household composition of four or more had statistically different proportions of respondents in each food security category, compared to high-income respondents with a household composition of less than four people. Of the 140 respondents who were high income and had four or more people in their household, 39% were considered food insecure, compared to only 11% of the 245 respondents who were high income but had less than four household members. The relationship between food insecurity and household composition is important to consider for households at all income levels, even high-income groups. According to the USDA Food and Nutrition Service (FNS), household size and income are taken into consideration when dispensing benefits. According to the FNS, households of four people must earn at most $2633 gross income a month to qualify, which is roughly $31,596 annually [27]. As shown in this analysis, even households of $75,000 or more can experience food insecurity, especially with four or more household members, but they may not qualify for benefits.

The impact of children on food security was reflective of the household composition findings. In all demographic comparisons and chronic health comparisons, households with children had higher instances of food insecurity and higher rates of more severe food insecurity than households without children. Children were also a significant variable in the logit estimations. This finding was supported by Coleman-Jensen et al. [11] who found that households with children had higher rates of food insecurity.

Comparable to the literature, chronic illness has an impactful relationship with food security. The unique finding in this study was that the results suggest that the nature of the illness may be important; however, it may be unrelated to food inherently. Diabetes was a significant indicator of food insecurity and was similar to the results found by Knight et al. [4]. The results of the cross-tabulations and logit estimations were reflective of Muldoon et al. [6]; generally, respondents with household depression/anxiety had higher rates of food insecurity. Diabetes, eating disorders, Crohn’s disease, celiac disease, high blood pressure, and high cholesterol all have food management considerations, but only diabetes and eating disorders were significant in the logit predictions. Depression/anxiety may be perceived to be the furthest removed from food and eating of the chronic illnesses investigated, but was identified as a significant variable. This suggests that the nature of food security and chronic illness may go beyond the intuitive relationship with food and that the specific nature of the illness may be important. As mentioned, aid programs were intended to support temporary conditions, but the results here suggest it is important to consider the impact of chronic conditions.

Another interesting result was the higher instance of food insecurity among males in the sample. This may seem counterintuitive to Coleman-Jensen et al. [11] who reported higher rates of food insecurity and very low food insecurity among households headed by single females. Marital status of the sample was not studied, although further analysis could shed light on this comparison. The results of this study were also similar to those found in Knight et al. [4] with higher rates of food insecurity among 18–44-year-olds.

This study also found that an individual was more likely to be food insecure as the poverty level of their county of residence increased. These results suggest that communities play a significant role in food security. These findings were not contradictory of the findings of Mayer et al. [10] and Kirkpatrick and Tarasuk [9], who suggest neighborhood-level impacts might not play a role in food security because both focused on grocery access. However, this was reflective of Dharamasen et al. [7] and Moore and Diez Roux [8] who considered regional income and poverty impacts. Overall food insecurity maybe related to limited food access due to resource constraints, both in the region and in the household, and not the specific availability of food markets.

Conclusion

It was found that 25% of the total sample were considered food insecure. Food insecurity was higher among males, middle-aged individuals, households with children, and low incomes. In the analysis of regional impacts, it was found that respondents living in counties with greater portions of the population living at or below the poverty line were more likely to be food insecure. When looking at chronic health, diabetes, eating disorders, and depression/anxiety were found to be significant variables predicting health score.

The researchers of this study acknowledge a number of health issues can be alleviated by improved food quality and access, so addressing chronic illness as a contributor of food security can improve the metrics of food security measurement and can improve policy designed to ease food access constraints on households. This study focused on the Midwest, but similar studies could be applied to measures around the world and inform global programs. As pointed out by Coates et al. [14], the CPS Food Security Supplement used by the USDA is comparable to a majority of measures used around the world. The Midwest represented a wide range of US food security, as measured by the CPS Food Security Supplement. The results reported here suggest that the CPS Food Security Supplement may underestimate the nature of food security by missing key factors such as income constraints (even in higher incomes), local community factors, and the impact of chronic illness. If the CPS Food Security Supplement underestimates US food security, measures similar to it around the world may also be underestimating.

While voter’s ballots and election polls suggest food security and health care are separate issues, it is clear from this study and a number of the studies reviewed, the two issues are related in significant ways. The current Farm Bill, the leading food and nutrition legislation in the USA, will be in effect until at least 2018 [28]. Establishing clear relationships between health and food can inform large pieces of legislation like the Farm Bill. The results presented here suggest the inclusion of chronic illness and health to improve metrics and inform food security legislation because of the impactful relationship between chronic illness and food security, specifically with respect to diabetes, eating disorders, and depression/anxiety.

Special consideration should also be given to high-income populations who may be ineligible under current policy for benefits but who could still be hungry due to household composition and household health conditions. Uncovering the impacts, especially impacts that were previously unknown or seem counter to convention (i.e., food insecurity challenges in high-income households), that current food policies have on Midwestern households is imperative in informing impactful future policies.

Policies evaluating food security should also include evaluations of the community. While this study focused on poverty indicators, reviewed studies have found other community variables that relate to food security. For effectiveness and efficiency, policy should consider all factors, case in point, policy addressing local food access by increasing food supply may not alleviate the issue if the target region is most impacted by income constraints related to limited health care. This study included children and household size, but other household dynamics such as marital status, the specific ages of the children, and number of generations in the household should be studied in future research.

Future studies of both health and food security could and should include an indicator for the other factor. Future research should also directly ask whether respondents actively chose between medication and food, and under what circumstances. It is also important to note that other illnesses require treatment that is not limited to medication and can lead to financial hardship, such as therapy. Future research should broaden the scope of illness and treatment and explore the impact on food security status. Additionally, future studies of health and food security should also include regional components in order to understand the impact of location on each. Policy makers should consider how food and health overlap when measuring food security and when generating programs to alleviate it.

Footnotes
1

For reference: for both the UDSA ten-question and eighteen-question surveys, respondents in the High Food Security group had a total score of zero and those considered food insecure had a score of three or higher.

 
2

For reference: a selection of “yes,” “often,” “sometimes,” “almost every month,” and “sometimes but not every month” contribute to increased food insecurity score.

 

Declarations

Authors’ contributions

SD participated in data collection, data analysis, and manuscript preparation. NJOW participated in data collection, data analysis, and manuscript preparation. AR participated in manuscript preparation. JZGW aided in data analysis. LA participated in data collection and manuscript preparation. All authors read and approved the final manuscript.

Acknowledgements

None.

Competing interests

The authors declare that they have no competing interests.

Availability of data and materials

The data sets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Consent for publication

Information on the research objective was provided to the survey respondents in advance of them opting into completing the survey instrument. Respondents were instructed of their ability to opt out of the survey at any point. The privacy and confidentiality of participants were also maintained.

Ethics approval and consent to participate

Approval for the data collection process employed for this analysis was obtained from the Purdue University Human Research Protection Program Institutional Review Board (IRB) under Protocol No. 1511016805.

Funding

This activity was funded, by Purdue, as part of AgSEED Crossroads funding to support Indiana’s Agriculture and Rural Development.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Authors’ Affiliations

(1)
Department of Agricultural Economics, Purdue University, West Lafayette, IN, USA
(2)
Department of Comparative Pathobiology, Purdue University, 725 Harrison Street, West Lafayette, IN 47907, USA
(3)
Department of Consumer Sciences, Purdue University, West Lafayette, IN, USA

References

  1. McCarthy N. The top issues for voters in the 2016 presidential election. Forbes. 2016. Accessed 2 Oct 2016 by S. R. Dominick from http://www.forbes.com/sites/niallmccarthy/2016/07/11/the-top-issues-for-voters-in-the-2016-presidential-election-infographic/#23cb2b508535.
  2. Office of the Press Secretary (OPS): The White Wouse. Remarks by the president at signing of the Farm Bill—MI. 2014. Michigan State University, East Landing Michigan. Accessed 2 Oct 2016 by S. R. Dominick from https://www.whitehouse.gov/the-press-office/2014/02/07/remarks-president-signing-farm-bill-mi.
  3. U.S. Department of Agriculture “Budget Summary and Annual Performance Plan: FY 2016.” 2015. S. R. Dominick. http://www.obpa.usda.gov/budsum/budget_summary.html. Accessed 10 Oct 2016.
  4. Knight CK, Probst JC, Liese AD, Sercy E, Jones SJ. Household food insecurity and medication “scrimping” among US adults with diabetes. Prev Med. 2016;83:41–5.View ArticlePubMedGoogle Scholar
  5. Berkowitz SA, Seligman HK, Choudhry NK. Treat or eat: food insecurity, cost-related medication underuse, and unmet needs. Am J Med. 2014;127:303–10.View ArticlePubMedGoogle Scholar
  6. Muldoon KA, Duff PK, Fielden S, Anema A. Food insufficiency is associated with psychiatric morbidity in a nationally representative study of mental illness among food insecure Canadians. Soc Psychiatry Psychiatr Epidemiol. 2013;48:795–803.View ArticlePubMedGoogle Scholar
  7. Dharamasen S, Bessler DA, Capps O Jr. Food environment in the United States as a complex economic system. Food Policy. 2016;61:163–75.View ArticleGoogle Scholar
  8. Moore LV, Diez Roux AV. Associations of neighborhood characteristics with the location and type of food stores. Am J Public Health. 2006;96(2):325–31.View ArticlePubMedPubMed CentralGoogle Scholar
  9. Kirkpatrick SI, Tarasuk V. Assessing the relevance of neighbourhood characteristics to the household food security of low-income Toronto families. Public Health Nutr. 2010;13(7):1139–48.View ArticlePubMedGoogle Scholar
  10. Mayer VL, Hillier A, Bachhuber MA, Long JA. Food insecurity, neighborhood food access, and food assistance in Philadelphia. J Urban Health Bull N Y Acad Med. 2014;91(6):1087–97.View ArticleGoogle Scholar
  11. Coleman-Jensen A, Rabbitt MP, Gregory C, Singh A. Household Food Security in the United States in 2014, ERR-194. U.S. Department of Agriculture, Economic Research Service; 2015.Google Scholar
  12. USDA Economic Research Service. U.S. Household Food Security Survey Module: Three-stage Design, with Screeners. 2012. Accessed 1 June 2016 by S. R. Dominick http://www.ers.usda.gov/datafiles/Food_Security_in_the_United_States/Food_Security_Survey_Modules/hh2012.pdf.
  13. Burchi F, De Muro P. From food availability to nutritional capabilities: advancing food security analysis. Food Policy. 2016;60:10–9.View ArticleGoogle Scholar
  14. Coates J, Frongillo EA, Rogers BL, Webb P, Wilde PE, Houser R. Commonalities in the experience of household food insecurity across cultures: what are measures missing? J Nutr. 2006;136:1438S–48S.View ArticlePubMedGoogle Scholar
  15. Hamilton WL, Cook JT, Thompson WW, Buron LF, Frongillo EA, Jr, Olson CM, Wehler CA. Household food security in the United States in 1995 technical report of the food security measurement project September. 1997. 1997 USDA Food and Consumer Service Office of Analysis and Evaluation 1–c1.Google Scholar
  16. Diansari P, Nanseki T. Perceived food security status—a case study of households in North Luwu, Indonesia. Nutr Food Sci. 2015;45(1):83–96.View ArticleGoogle Scholar
  17. Headey D, Ecker O. Rethinking the measurement of food security: from first principles to best practice. Food Secur. 2013;5:327–43.View ArticleGoogle Scholar
  18. U.S. Census Bureau; American Community Survey, 2014 American Community Survey 1-Year Estimates, Table S1901; generated by S. R. Dominick; using American FactFinder; http://factfinder2.census.gov. 21 Sept 2015.
  19. U.S. Census Bureau (USCB). Geography “Geographic Terms and Concepts—Census Divisions and Census Regions”. 2015. Accessed 8 Aug 2015 by S. R. Dominick from https://www.census.gov/geo/reference/gtc/gtc_census_divreg.html.
  20. U.S. Census Bureau (USCB 2). State and County Estimates for 2015. 2015. Accessed 8 Aug 2015 by S. R. Dominick from http://www.census.gov/did/www/saipe/data/statecounty/data/2015.html.
  21. USDA Economic Research Service. U.S. household food security survey module: three-stage design, with screeners. 2012. Accessed 1 June 2016 by S. R. Dominick http://www.ers.usda.gov/datafiles/Food_Security_in_the_United_States/Food_Security_Survey_Modules/hh2012.pdf.
  22. IBM Corp. Released 2015. IBM SPSS Statistics for Windows, Version 23.0. Armonk: IBM Corp.Google Scholar
  23. StataCorp. Stata statistical software: release 14. College Station: StataCorp LP; 2015.Google Scholar
  24. Migliore G, Schifani G, Cembalo L. Opening the black box of food quality in the short supply chain: effects of conventions of quality on consumer choice. Food Qual Prefer. 2015;39:141–6.View ArticleGoogle Scholar
  25. Peterson HH, Taylor MR, Baudouin Q. Preferences of locavores favoring community supported agriculture in the United States and France. Ecol Econ. 2015;119:64–73.View ArticleGoogle Scholar
  26. Baum CF. An introduction to modern econometrics using Stata. College Station: Stata Press; 2006.Google Scholar
  27. USDA FSN Food and Nutrition Service. “Supplemental Nutrition Assistance Program (SNAP): Eligibility. 2016. Accessed 12 Dec 2016 by S. R. Dominick from http://www.fns.usda.gov/snap/eligibility#Income.
  28. USDA Economic Research Service. Agricultural Act of 2014: highlights and implications. 2016. Accessed 1 Nov 2016 by S. R. Dominick https://www.ers.usda.gov/agricultural-act-of-2014-highlights-and-implications/.

Copyright

© The Author(s) 2018

Advertisement