Food insecurity in US households with children with limiting health conditions was compared to households with children without limiting health conditions, controlling for demographic variables. Multivariable logistic regression was used to calculate the odds ratios. Data from the 2008-2011 Making Connections Survey (<i>N</i> = 1940) of households with children in seven high poverty communities in the US were used. Having a child with a limiting health condition made a household’s odds 1.41 times (95% C.I., 1.110, 1.790) more likely to be food insecure. When there are two or more children with limiting conditions in the household, the odds of food insecurity are 1.67 times (95% CI, 1.16, 2.40), higher than a family with no children with a disabling health condition. Families with children with limiting health conditions in high poverty communities are especially at risk of experiencing food insecurity, which can complicate health conditions. Nutrition assistance programs are vital to keep children and families food secure.
One of out every six households in the US is food insecure (defined as limited or uncertain availability of nutritionally adequate and safe foods) with nearly a quarter (24.5%) of all children under the age of 6 living in food insecure households in 2011 [
Families with children with limiting health conditions (physical, learning, mental, or chronic health conditions that limit participation in the usual kinds of activities done by most children his/her age) can be especially susceptible to family financial hardship due to the increased direct (service needs) and indirect costs of disability [
This article explores the question of how food insecurity differs for families with children with limiting conditions and those without in data collected during and after the severe economic downturn (2008-2011). This research analyzes data from the Annie E. Casey Foundations’, Making Connections Project, which collected data from 4300 households across seven high poverty communities focusing on disadvantaged neighborhoods.
In 2013, the Food Stamp Program experienced its largest funding decrease in history with additional substantial decreases to this program coming in the following years as a result of the passed farm bill [
Very limited prior research on this topic exists and uses National Survey of America’s Families data collected between 2000 and 2002 or Survey of Income and Program Participation data collected between 2004 and 2008 [
This research evaluates whether the odds of being food insecure are different for households with children who have a limiting health condition compared to households with children who do not have a limiting health condition, after controlling for relevant demographic variables. A second analysis evaluates if having more than one child in the household with a limiting health condition changes these odds.
A cross section (Wave 3) of data from the longitudinal making connections dataset was used for this study. Data were collected between 2008 and 2011 in seven high poverty communities located in the following metropolitan areas: Des Moines, IA; Indianapolis, IN; Denver, CO; San Antonio, TX; Seattle, WA; Providence, RI; and Louisville, KY. The seven communities were chosen in order to represent the different geographic areas of the United States but were all economically disadvantaged communities. The reasons for the economic disadvantages included: declining neighborhoods in older industrial cities (Louisville, Milwaukee, Indianapolis), expand- ing immigrant populations (Des Moines, Hartford, Providence) and racially diverse neighborhoods with a severe shortage of affordable housing (San Antonio, Denver, Oakland, Seattle). Area probability sampling via US postal addresses was used to select a random sample of addresses (N = 4315 households) within the targeted neighborhoods. One adult respondent was selected to provide information about themselves, their spouse or partner (if applicable), and any children between the ages of 0 and 17 living in their residence. Response rates ranged from 75% to 87% among the seven sites. Households without children (N = 2375) were removed from analysis, leaving a total of N = 1940 households in the analytic dataset.
Informed consent was obtained by the University of Chicago’s NORC team of researchers that conducted the survey. The dataset has been stripped of identifying information so that researchers who conduct secondary data analysis are unable to make re-identification of the respondents. The authors of this paper obtained IRB approval from their perspective universities before analyzing the data. The Annie E. Casey Foundation requires that researchers go through an application and approval process before accessing the dataset including completing training on the survey and its’ data. All authors of this paper underwent that process successfully.
Respondents were administered a scripted interview which was recorded on a paper and pencil questionnaire via in-person or telephone interview. Interview surveys were available in English, Spanish, and other languages that at least 10% of the local population spoke.
The dependent variable, food insecurity, was measured by asking the adult respondent “In the last 12 months… was your family ever without enough money to buy food.” Four respondents who answered “don’t know” or “refused” were removed leaving a maximum of N = 1936 households for logistic analysis. Control variables measured include adult respondent’s gender (84.4% = female), adult respondent’s age in years (median = 35 years, standard deviation 11.2 years), adult respondent’s self-reported race (Non-White, 70.5%, White 29.5%), and presence of a spouse or partner who lives in the household (yes = 11.7%). Adult respondent’s highest level of education attained was used as a proxy for household income since 27% of the values are missing for the income variable which disqualifies it as a candidate for missing data imputation. Education only has 26 missing values and was measured using a nine category ordinal variable which ranged from “eighth grade or less” to “graduate degree.” The education variable’s fit as a proxy was assessed by comparing observations where both variables were available, and it was found that the mean and median total household income increased as education category increased, confirming a strong positive relationship between education and total household income. One widely cited limitation of using education, as a proxy for income is that income’s relationship to education varies by race and gender [
To measure the independent variable, presence of a limiting condition, respondents were asked for each child living in the household, “Has a health professional ever told you that your child has a physical, learning, mental, or chronic health condition that limits his or her participation in the usual kinds of activities done by most children his or her age or limits his or her ability to do regular school work?” This variable was coded ordinally (0 = no children with a disabling health condition 75.9%, 1 = only one child with a disabling health condition 19.1%; 2 = two or more children with a disabling health condition 5.0%).
To view the entire questionnaire go to: http://mcstudy.norc.org/documentation/ and to see the sample selection process go to http://mcstudy.norc.org/study-design/files/MkCon%20Sample%20Design.pdf.
Chi-square analyses were run to assess the bivariate relationships between each demographic variable and whether or not the household is food insecure. Then mulitvariable logistic regression was used to calculate odds ratios with 95% confidence intervals to describe the relationship between having a child with a limiting health condition and food insecurity while controlling for relevant demographic variables. A second mulitivariable model was constructed to further define the odds of experiencing food insecurity for families that have zero, one, or more than one, child with a limiting condition.
Chi square analysis revealed that food insecurity is related to all measured explanatory variables except having a spouse or partner who is present in the household (see
Demographics | All households | Food insecure households | Food secure households | p-value | |||
---|---|---|---|---|---|---|---|
n | % | n | % | n | % | ||
Food security (missing information on n = 4) | 1936 | 521 | 26.9% | 1415 | 73.1% | ||
Respondent sex | 0.0014 | ||||||
Male | 303 | 15.6% | 59 | 19.5% | 244 | 80.5% | |
Female | 1635 | 84.4% | 462 | 28.3% | 1169 | 71.5% | |
Missing information | 2 | ||||||
Respondent race | 0.0486 | ||||||
Non-White | 1345 | 70.5% | 378 | 28.1% | 965 | 71.7% | |
White | 562 | 29.5% | 133 | 23.7% | 427 | 76.0% | |
Missing information | 33 | ||||||
Is respondent’s spouse or partner present | 0.4105 | ||||||
No | 1680 | 88.3% | 459 | 27.3% | 1217 | 72.4% | |
Yes | 222 | 11.7% | 55 | 24.8% | 167 | 75.2% | |
Missing information | 38 | ||||||
Household receiving food stamps (SNAP) | <0.0001 | ||||||
No | 949 | 49.4% | 197 | 20.8% | 752 | 79.2% | |
Yes | 973 | 50.6% | 319 | 32.8% | 650 | 66.8% | |
Missing information | 18 | ||||||
Respondent education | 0.0011 | ||||||
Eighth grade or less | 207 | 10.8% | 66 | 31.9% | 140 | 67.6% | |
Beyond eighth but not high school graduate | 417 | 21.8% | 132 | 31.7% | 284 | 68.1% | |
GED—general education diploma | 150 | 7.8% | 49 | 32.7% | 101 | 67.3% | |
High school graduation | 460 | 24.0% | 117 | 25.4% | 342 | 74.3% | |
Trade or vocational school | 99 | 5.2% | 28 | 28.3% | 71 | 71.7% | |
One to three years of college | 424 | 22.2% | 99 | 23.3% | 324 | 76.4% | |
Graduated four year college | 94 | 4.9% | 15 | 16.0% | 79 | 84.0% | |
Some graduate education | 17 | 0.9% | *ND* | *ND* | |||
Graduate degree | 46 | 2.4% | *ND* | *ND* | |||
Missing information | 26 | ||||||
Any child with limiting condition | 0.0002 | ||||||
No | 1473 | 75.9% | 365 | 24.8% | 1105 | 75.0% | |
Yes | 467 | 24.1% | 156 | 33.4% | 310 | 66.4% | |
Number of children with limiting condition | 0.0007 | ||||||
0 | 1473 | 75.9% | 365 | 24.8% | 1105 | 75.0% | |
1 | 371 | 19.1% | 120 | 32.3% | 250 | 67.4% | |
2+ | 96 | 5.0% | 36 | 37.5% | 60 | 62.5% |
n | Median | n | Median | n | Median | p-value | |
---|---|---|---|---|---|---|---|
Respondent age (yr) | 1937 | 35 | 519 | 36 | 1414 | 35 | 0.0283 |
Total household income ($000) | 1413 | 22 | 395 | 17 | 1016 | 25 | 0.0001 |
Number of adults in household | 1940 | 2 | 521 | 2 | 1415 | 2 | 0.0012 |
Number of children in household | 1940 | 2 | 521 | 2 | 1415 | 2 | 0.0010 |
*ND* = not provided due to the disclosive nature of the information as required by Annie E. Casey Foundation.
between each of the variables and food insecurity. The only variables that were not related were food insecurity and whether the respondent had a spouse or partner present in the household. This may be due to the fact that 88% of households in this sample were single parent households.
Multivariable logistic regression analysis of the variables measured found (see
A second model evaluated if having more than one child in the household with a limiting health condition would change these odds. The odds ratios for all the control variables were nearly identical in both models. The odds of a household being food insecure when they have one child with a limiting condition versus no child with a limiting condition is 1.29 (95% CI, 1.08 - 1.55). When there are two or more children with limiting conditions in the household, the odds of food insecurity are 1.67 (95% CI, 1.16 - 2.40) times higher than a family with no children with limiting health conditions (see
Families with children with limiting health conditions have an array of added direct and indirect costs due to their children’s health conditions. Direct costs include health care expenses, copays, deductibles, fees to see specialists, medical equipment, prescription drugs and therapy services. Indirect costs can include additional child care expenses, housing renovations, and educational expenses. Indirect costs can also include loss of income that parents can experience from having to miss work in order to care for children or to take children to doctors and therapy appointments. Women especially face these challenges as their employment has been found to be more impacted than men’s employment by the presence of children with limiting health conditions [
There are some limitations to this research that must be noted. The first being that this study is not able to establish a causal relationship between children’s limiting health conditions and food insecurity. Correspondingly, the direction of the relationship between children’s health status and food insecurity in this cross-sectional data can- not be established. Second, although random sampling was used within the areas studied, the seven metropolitan areas themselves were selected in a non-random fashion, which limits the generalizability of these findings. Another limitation is that the data relies upon caregiver reports of both food insecurity and child’s health status. Further research that may be able to clarify the extent of both variables is needed. The strength of this research is its focus on recent data during the economic recession. The importance of nutrition programs in supporting financially and health fragile families is high.
Future research should expand this inquiry to investigate the role of insurance status, receipt of coordinated care, type and severity of health condition, and receipt of government and employment benefits as they relate to
Multivariable results-model 1 | OR | CI | p-value | |
---|---|---|---|---|
Respondent’s age (years) | 1.016 | 1.006 | 1.026 | 0.0013 |
Respondent’s sex (female = 1) | 1.576 | 1.134 | 2.189 | 0.0067 |
Respondent’s race (White = 1) | 0.821 | 0.646 | 1.043 | 0.1069 |
Respondent spouse or partner (present = 1) | 1.008 | 0.718 | 1.416 | 0.9623 |
Respondent education | 0.906 | 0.857 | 0.957 | 0.0004 |
Food stamps (yes = 1) | 1.769 | 1.414 | 2.215 | 0.0001 |
Any child with a limiting conditions (yes = 1) | 1.409 | 1.110 | 1.790 | 0.0049 |
n = 1823; c statistic = 0.634.
Multivariable results-model 2 | OR | CI | p-value | |
---|---|---|---|---|
Respondent’s age (years) | 1.016 | 1.006 | 1.026 | 0.0014 |
Respondent’s sex (female = 1) | 1.578 | 1.136 | 2.193 | 0.0065 |
Respondent’s race (White = 1) | 0.822 | 0.647 | 1.045 | 0.1091 |
Respondent spouse or partner (present = 1) | 1.017 | 0.724 | 1.428 | 0.923 |
Respondent education | 0.906 | 0.858 | 0.958 | 0.0005 |
Food stamps (yes = 1) | 1.762 | 1.407 | 2.206 | 0.0001 |
Number of children with limiting condition (0, 1, 2+) | 1.293 | 1.079 | 1.549 | 0.0053 |
n = 1823; c statistic = 0.635.
food insecurity and children’s health status. The health benefits of systematic assessment of food insecurity, and when appropriate, referral for those most at risk is another important area for further study.
Clinicians caring for children with limiting health conditions must take a holistic approach to improving the lives of their patients. Not only are children with limiting health conditions sick they may be hungry too. This research suggest that clinicians may need to screen for food insecurity with parents and older children if possible bearing in mind that discussing hunger can be culturally complex and shameful [
A second clinical model found to support families with children with limiting health conditions is to provide care in a medical home setting. Researchers have consistently found that receiving care in a medical home safeguards parental employment, which can mean less financial instability [
There are a number of federal nutrition programs in the United States that may be able to assist food insecure households including the Supplemental Nutrition and Assistance Program (known as Food Stamps), Women, Infant, and Children (WIC), School Food Programs, and summer school feeding programs. The largest historic funding decrease to the Food Stamp program (totaling $5 billion for fiscal year 2014) went into effect in November of 2013. On average families (of 3) lost $29 a month ($36 a month for a family of 4) in benefits bringing down the average benefit to $1.40 per person per meal [
Parents of children with limiting health conditions need expanded access to short-term family leave programs. The federal Family and Medical Leave Act only provides for leave on an unpaid extended time frame of up to 12 weeks. What parents of children with limiting health care conditions really need access to is short-term protected (preferably paid) sick leave that allows them to take children to doctor’s appointments or to be off work for a day or two at a time when children need care. In 2006, San Francisco became the first municipality to guarantee paid sick leave to all workers [
In conclusion, families with children with limiting health conditions are more likely to be food insecure and families with more than one child with a health condition are even more at risk. Researchers, clinicians and policy makers must work together to support these families nutritionally and financially so that further health conditions are not exacerbated.
None of the authors have any conflicts of interest relevant to this article.