National Poverty Center Gerald R. Ford School of Public Policy, University of Michigan www.npc.umich.edu
Conference on Poverty and Health This paper was delivered at a National Poverty Center conference. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the view of the National Poverty Center or any sponsoring agency. Impact of Poverty and Household Food Security on the Use of Preventive Medical
Services in the California Health Interview Survey
Gail G. Harrison
UCLA School of Public Health and
UCLA Center for Health Policy Research
Los Angeles, CA
gailh@ucla.edu
Paper presented for discussion at the National Poverty Center Conference on Poverty and
Health, Ann Arbor, Michigan, July 20-21, 2004
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Acknowledgements
I thank the following contributors:
E. Richard Brown, who is Principal Investigator for the California Health
Interview survey
Charles DiSogra, who is Director of the California Health Interview Survey
Wei Yen, PhD and Lida Becerra, MS, for programming assistance
This work was supported by small grants from USDA/Economic Research Service
through the UC Davis Small Grants Program of the Food Assistance and Nutrition
Research Program, and from the National Poverty Center, University of Michigan.
2
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Table of Contents
Background
Definition and Measurement of Household Food Insecurity
Health Consequence of Household Food Insecurity
Research Questions
Methods
The California Health Interview Survey
Analytical Methods
Page
4
6
6
7
9
Results
Prevalence and Predictors of Household Food Insecurity
9
Use of Preventive Medical Services by Food Security Status
10
Use of Medical Care by Food Security Status
11
Among Adults with Chronic Disease, Postponement or Foregoing
of Prescribed Care by Food Security Status
12
Food Stamp Program Participation
16
Discussion
16
References Cited
Appendix I: Harrison, DiSogra, Manalo-LeClair, Aguayo, Yen. Over 2.2 Million LowIncome California Adults are Food Insecure; 658,000 Suffer Hunger. UCLA Health
Policy Brief, November 2002.
Appendix II: Multivariate Prediction Models
3
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Introduction and Background
Household Food Insecurity: History and Measurement in the US
The concept of food security originated in the international development literature
of the 1970s, and was originally utilized on an aggregate (country or regional) level to
refer to the ability of a sociopolitical entity to produce or import sufficient food supply to
maintain its population. The term “hunger” has long been used by human rights and
public health advocacy groups to refer to inadequate access to food at the individual or
household level, but has been variably and usually ill-defined. Beginning in the 1980s,
there began to be serious attention paid to the measurement of hunger in the North
American context, through two lines of work. The Community Childhood Hunger
Identification Project (CCHIP), an advocacy effort, conducted surveys in several states in
which direct interview questions about experiences and coping strategies among mothers
of young children resulted in some validation of constructs about management strategies
in dealing with lack of adequate access to food. At the same time a group centered at
Cornell, beginning with Radimer’s dissertation, undertook systematic formative work to
explore the “managed process” of hunger – detailed qualitative work undertaken
exclusively in the Northeast, focusing on the psychological and experiential precursors to
hunger, a domain that has since been labeled “food insecurity.”
In 1990, two events occurred that resulted in moving forward rapidly in the area
of measurement of household food insecurity. One was the LSRO/FASEB consensus
panel that articulated an operational definition of the concept of food insecurity as lack of
continuous, secure access at all times to a diet adequate to support healthy life and 1) the ready
availability of nutritionally adequate and safe foods and 2) the assured ability to acquire
personally acceptable foods in socially acceptable ways (Hamilton 1997). The second was
the passage of the National Nutrition Monitoring and Related Research Act of 1990,
which mandated that USDA and DHHS develop and validate an appropriate instrume nt
for the epidemiologic measurement and monitoring of the prevalence of food insecurity
and hunger in the US population.
The United States Food Security Instrument, developed in the early 1990s and
used since 1995 to monitor prevalence of food insecurity and hunger through the Current
Population Survey, consists of 18 questions that deal with various aspects of household
food insecurity. The questions represent a range of food insecurity conditions, beginning
with questions on the inadequacy of food supplies and money available for food. Worry
and concern about having adequate amounts of food are also included in the beginning of
the scale. As participants respond to the questionnaire they move to questions that
indicate reduced food intake for adults and finally for children (Cohen et al.,1999). The
behaviors that the questions refer to generally occur in an ordered sequence as the
severity of food insecurity increases. Adults in the household typically worry about
having enough food, then stretch household resources and juggle other necessities such as
utility bills or rent. They then tend to decrease the quality and variety of household
4
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
members’ diets, and then decrease the frequency and quantity of adults’ food intake.
Finally, a decrease in the frequency and quantity of children’s food intake occurs (Nord,
Jemison and Bickel, 1999, Hamilton et al., 1997). The concept is to be distinguished
from food insufficiency, a less conservative measure that was used in numerous national
surveys prior to the development of the current instrument.
Cutoff points for the US Food Security Survey Instrument have been determined
that place respondents into one of four categories:
Food Secure: Households show no or minimal evidence of food insecurity.
Food Insecurity with No Hunger Evident: Food Insecurity is evident in household
concern about adequacy of household supply and the adjustments made by the
household in managing their supplies, including reducing the quality of food and
an increase in unusual coping patterns. There is little or no reduction in household
member’s intake.
Food Insecurity with Hunger Evident: Adults in the household have reduced their
food intake to an extent that implies that they have repeated experiences with the
physical sensation of hunger.
Food Insecurity with Severe Hunger Evident: For Households with children, this
level implies that the children’s food intake has been to an extent that implies that
the children have repeated experiences with the physical sensation of hunger. For
households without children and for some adults living in households with
children, this level implies a more severe level of household hunger.
The US Household Food Security Instrument is now embedded in a number of
national surveys including the National Health and Nutrition Examination Survey. A
standardized Spanish translation has been developed and published (Harrison et al., 2002);
a shortened six- item screener has been validated (Blumberg et al., 1999); and adaptations
are being developed and applied in a wide variety of populations in the US and other
countries.
Data from 2002 indicate that about 11.1% of US households were food insecure
by this measure, including 3.5% who experienced hunger (up from 10.7% and 3.3%
respectively in 2001 (Nord et al., 2002). The relationship of household food insecurity to
poverty measured by per capita income is close but not exact; about one-third of
households below the federal poverty line are classified as food insecure, compared to
about 8% of households with incomes above the poverty line. While food insecurity is an
index of severe economic stress, there is inevitably a management component as well.
Subgroups of the population with higher prevalence of food insecurity than the national
average are Hispanic and African-American households and households composed of a
single adult and children.
5
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Health Consequences of Household Food Insecurity.
Adverse consequences of resource limitations so severe as to result in food
insecurity include not only compromise dietary quality and nutritional status but also
result in detrimental outcomes not mediated through nutritional status. It has been
demonstrated in several U.S. subpopulations and in national US and Canadian samples
that hunger or risk of hunger is directly linked to poor physical, social, and mental well
being and to a decreased quality of life in a variety of populations including adults,
adolescents, and young children (Frongillo et al., 1999, Rose, 2000; Alaimo et al. 2001a,
2001b, 2002; Nicholas et al., 2003; Tarasuk 2001, Siebert et al., 2001; Kleinman et al.,
1998; Stormer and Harrison 2003).
For the one-third of the adult population that has one or more chronic illnesses
that require ongoing management including drugs, diet and lifestyle variables, there are
other potential adverse health consequences. Doctor visits, prescription drugs and
devices, and special diets all require ongoing monetary outlays and also have indirect
costs. Even with health insurance there are deductibles, co-payments, child care and
transportation costs and opportunity costs. For the individual with chronic illness,
inability to meet these needs due to the more immediate press of putting food on the table
may result in poorer disease management, increased complication rates, and adverse
outcomes. We and others have shown food insufficiency to be related to poorer disease
management, poorer health status and increased health care utilization for low income
adults with diabetes (Nelson et al., 1998, Nelson et al., 2001), an observation that
produced the impetus for the present analysis.
The Research Questions.
We asked the question of whether, for individuals in low-income households
facing food insecurity, a tradeoff between medical care and food could be demonstrated,
and if so, what the effects of health insurance and of food assistance program
participation (particularly food stamps) might have on that tradeoff. Specifically, we
hypothesized that:
1) Low- income adults in households that exhibit food insecurity will be more
likely than those in food-secure households to report non-use of preventive
medical services.
2) In the presence of diagnosed chronic disease, adults in food insecure
households will be more likely than those in food secure households to report
postponement of filling prescriptions for prescribed drugs and failure to follow up
or delay in following up on other recommended medical care.
3) Health insurance will at least partially mitigate these hypothesized effects.
6
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
4) Partic ipation in food assistance programs will mitigate the hypothesized
relationships mentioned above, independent of health insurance status.
Methods
The first round (2001) of the California Health Interview Survey (CHIS) was the
data set for the current analysis. CHIS is conducted by the UCLA School of Public
Health in cooperation with the California Department of Health Services and the Public
Health Institute in Berkeley, and is the largest state health survey conducted in the United
States. The first round in 2001 collected information from 55,428 households, drawn
from every county in the state. Individual interviews were completed for one adult per
household and from one adolescent (aged 12-17) and with a parent on behalf of one child
under 11 years when these were present in the household, resulting in 55,428 adult
interviews, 5801 adolescents and 12,592 parents about a child under 11 years. CHIS is a
telephone-based survey, designed to be implemented every two years, and is conducted
in six languages (English, Spanish, Mandarin, Vietnamese, Hmong, and Korean). It
includes information on health status, specific conditions and disease management, health
behavior, women’s health, oral health, mental health, cancer screening history, health
care access and utilization, health insurance, food security, physical activity, body weight,
some dietary information and public program participation. CHIS 2001 was funded by
the California Department of Health Services, the California Endowment, the California
Children and Families Commission, the National Cancer Institute, the Centers for
Disease Control and Prevention, and the Indian Health Service. The second round of
CHIS (2003) is now in the data-cleaning process, and planning is underway for CHIS
2005.
Sampling Design. The CHIS sample is designed for two purposes: to produce
statewide prevalence estimates for California’s overall population and its major ethnic
groups, including some ethnic subgroups; and to provide local- level estimates for
counties with populations of 40,000 or more, for local planning and to enable
comparisons among counties. The basic state-wide sample was selected through a
random-digit dial process; only residential households with telephones are in the CHIS
sample, but results will be statistically adjusted to account for households without
telephones. Adult sample sizes by ethnic group from the random digit dial sample were:
34,383 non-Latino white, 4377 Latino, 2615 African-American, 3928 Asian American,
931 Native American, 236 Native Hawaiian or Pacific Islander, and 8954 other and
mixed. CHIS incorporates over-samples of some geographic areas and some ethnic
groups. The three cities in California with their own health departments were oversampled to reach a target sample size of 800 per city. Also, because Asian subgroups are
heterogeneous in culture and language, an additional 2100 households were surveyed in
five specific Asian subgroups that would otherwise have samples too small for adequate
estimates. Finally, American Indian/Alaska Natives were over-sampled to raise their
total sample size to at last 800 in order to examine differences between rural and urban
areas. All 58 California counties were included in the sample design, arranged into 41
strata. Thirty-three of the 35 counties with a population of 100,000 or more form their
7
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
own sample strata; two are combined with smaller adjoining counties, and the remaining
23 counties are grouped into six sample strata in a way that is meaningful for health
planning purposes. The minimum target sample size from any stratum is 800. Within a
household, up to three individuals were part of the survey: one adult (18 years or older),
plus in households with adolescents (12-17 years) residing at home one was included, and
in households with children 11 years and under, the adult most knowledgeable about the
child was interviewed about that child.
Survey methods. CHIS is a telephone-based, computer-assisted survey modeled
largely on the National Health Interview Survey, with extensive local adaptation.
Extensive pre-survey publicity, call-back protocols, and CHIS’ multilingual capability all
contributed to a relatively good response rate, with 59.2% of sampled phone numbers
resulting in completion of the screener, and of those 63.7% completing the extended
interview (overall response rate, weighted by probability of being sampled, 37.7%).
(California Health Interview Survey, 2002). The entire interview required between 25
minutes and one hour, depending mostly on household size and partly on language of the
interview.
Key variables for the present analysis. Of interest for this analysis, household
income was asked in categories and then within the CATI protocol per capita income as
percent of the Federal Poverty Level (FPL) calculated and categorized as <100% FPL,
100-199% FPL, 200-299% FPL, and 300% FPL+. In order to examine the effects of
public program participation, we imputed cutoffs at 130% FPL, the income eligibility
criterion in most states for food stamp eligibility, and 185%, the cutoff for MediCal
health insurance eligibility as well as for the WIC program. The food security
measurement was the six- item screener based on the US Food Security Instrument
(Blumberg et al.,1999) and was asked only of adults in households whose per capita
income was calculated to be <200% of the FPL.
Use of medical care was explored with the questions relating to number of doctor
visits in the previous 12 months, time since last dental visit, visits to mental health care
providers and to other providers in the previous 12 months, and # of emergency room
visits in the last 12 months both generally and for complications of any specific chronic
disease. The presence of chronic disease was ascertained by asking “Has a doctor ever
told you that you have (name of disease)?”. For purposes of the present analyses, we
utilized data from adults who responded positively that they had any of five specific
conditions, namely arthritis, asthma, heart disease, high blood pressure, and diabetes.
These were selected because they make ongoing management demands in terms of
compliance with prescribed drug regimens and periodic re-examination. We treated
diabetics taking oral drugs separately from diabetics taking insulin because of the
differential demands for self-care and the differential tendency for acute complications.
Disease management was indexed by a series questions about whether in the last 12
months the individual had failed to follow up or delayed in following up in getting a
prescription filled (both in general and for his/her specific condition) and in getting
medical tests or other care (including referrals to specialty care and non- medical
providers such as physical therapists and dietitians). Positive responses to these
questions were followed by a question about “What was the reason for that?” that
8
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
included “could not afford it” as one of several options. Health insurance was explored
in considerable detail in CHIS; in the present analyses we use a simple dichotomous
variable (do or do not currently have any health insurance). Data on public program
participation are also present in CHIS; for the present analysis we utilized data on food
stamp program participation only; the numbers of adults with chronic disease that we
could assume were eligible for the WIC program were too small for stable multivariate
prediction models.
Analytical methods. SUDAAN was used to apply sample weights to account
design effects; analyses were accomplished in SAS Version 8.0.2. Basic bivariate
relationships were examined for key variables; multivariate predictions of dichotomous
dependent variables were accomplished through multiple logistic regression analyses.
Potential interactions were explored both by the construction of interaction terms and by
moving key variables in and out of the models to explore the effects on coefficients for
other candidate variables.
Results
Prevalence and Predictors of Household Food Insecurity.
Table 1 shows the prevalence of household food insecurity without and with
hunger for the low-income (<200% FPL) portion of the CHIS adult sample. Overall,
28.3% reported food insecurity. As expected, the prevalence of both total food
insecurity and the more severe food insecurity with hunger were higher among those in
households with incomes blow the poverty line than for those at 1 – 2 x the poverty level.
The descriptive data on prevalence of food insecurity have been published as a UCLA
Health Policy Brief (Harrison et al., 2002), a copy of which is appended to this paper
(Appendix 1). Briefly, the population subgroups with the highest prevalences were
American Indian and Alaska Natives, Hispanics, and African Americans; households
consisting of a single adult with children; and households in California’s Central Valley
and the rural northern part of the state. Fewer than 20 percent of income-eligible adults
who reported food insecurity with hunger were participating in the food stamp program.
9
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Table 1
Prevalence (%) of Househo ld Food Insecurity among Low-Income California Adults
(<200% of Federal Poverty Line), 2001
Food Security Status
Household Income
0-99% Poverty 100-199% Poverty
Food Secure
63.8
77.3
Food Insecure without Hunger
25.2
16.4
Food Insecure with Hunger
11.5
6.4
We explored in multivariate models several potential predictors of household food
insecurity (Appendix II- i) and of food insecurity with hunger (Appendix II- ii). Briefly,
food insecurity was predicted by income within the low-income range, with lower risk for
individuals in households above than below the poverty level, with the exception of those
above the income-eligibility cutoff for MediCal (185% FPL) whose risk was equal to
those below the poverty level. Individuals ages 36-54 were more likely to be food
insecure than those below 35 years; adults older than 65 were significantly less likely to
report food insecurity than younger adults, consistent with our earlier bivariate
observations (Appendix I). Latinos, African Americans and American Indian/Alaska
Natives were more likely than white or AAPI adults to be food insecure. Adults in
families with children, particularly when headed by a single adult, and those without
health insurance were more likely than others to be food insecure. The presence of a
chronic illness did not independently predict food insecurity; on the contrary, a single
chronic illness for some reason was negatively predictive, other things being equal.
Prediction of food insecurity with hunger showed similar relationships with age,
household income, family type and chronic illness. The relationships with ethnicity were
different, with Latino and AAPI adults significantly less likely than those of other
ethnicities to report household- level hunger. There was no relationship between health
insurance status and food insecurity with hunger in the multivariate model. Additionally
we ran models with additional indicators of ill health (general self-rated health as fair or
poor, functional limitations due to health reasons, overnight hospital stay in last year,
taking medications); none had significant predictive value for food insecurity or food
insecurity with hunger.
Relationships between household food insecurity and use of preventive medical
services.
Table 2 shows the percentage of low- income adults reporting utilization of
various types of preventive health services, by food security status. There was no
relationship for several services, including having had a flu shot in the last year, cancer
10
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
screening (mammograms, Pap smears) for women, and having had fecal occult blood
testing as screening for colon cancer. There was also no effect on the probability of
having a “medical home” – i.e., a place to which the individual reports usually going
when needing medical care or health advice. Only about 60 percent of low- income
individuals reported having a “medical home” regardless of food security status. There
were slight differences (lower section of Table 2) in prevalence of ever having had a
bone density test for older women, ever having had a PSA test for prostate cancer
screening for men >40, ever having had an endoscopic colorectal cancer screening, ever
having had a blood cholesterol check (although most individuals did report having had
this at some time), and in having had more than five years elapse since the last dental
visit.
Table 2
Percentage of Low-Income Adults Who Reported Use of Various Preventive Health
Services, by Household Food Security Status
Food Secure Food Insecure
w/o hunger
Have a medical home
59.6
57.5
Had a flu shot in the last year
65.2
60.5
Ever did a stool blood test
46.5
37.5
Ever had a mammogram (women 40+) 86.1
82.3
Most recent mammogram>5 yrs ago
6.1
5.0
Ever had a Pap smear (women 18+)
89.1
90.2
Ever had bone density test (women 50+) 29.4
Ever had a colon/rectal exam
47.7
Ever had a PSA test (men 40+)
56.3
Ever had a blood cholesterol check
96.1
Last dental visit >5 years ago
11.3
22.8
38.9
36.5
99.9
13.1
Food Insecure
with hunger
65.0
63.2
41.2
83.9
7.8
94.6
22.9
37.8
49.1
79.4
17.6
Use of medical care by food security status
Examination of the data on number of physician encounters and use of emergency
rooms over the year prior to the survey, as well as use of mental health care providers,
showed greater utilization by individuals in severely food insecure households (with
hunger) than by others. Adults in households reporting food insecurity with hunger were
approximately twice as likely as food-secure low- income adults to have had seven or
more doctor visits in the previous year and to have visited an emergency room for care in
the same period. They were more than twice as likely to have seen a mental health care
provider. There was a mild apparent effect of milder levels of food insecurity (without
hunger) (see Table 3). For individuals in food-secure low- income households, utilization
was not different from those in higher- income (>200% FPL) households (data not shown).
11
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Table 3
Utilization of Physician Care, Emergency Room care, and Mental Health Care
in 12 Months Prior to the Survey
>7 doctor visits
in previous yr
>1 ER visit
in previous yr
Visited mental health
provider in previous yr
Food Secure
13.7
14.4
3.8
Food Insecure
without Hunger
16.7
17.8
6.3
Food Insecure
with Hunger
26.8
26.6
9.8
Among adults with chronic diseases, risk of foregoing or postponing needed care
by food security status
Almost one-third of all adults report having one or more of the chronic diseases
we identified as candidates for this analysis. The proportion for most conditions was
marginally higher in the low- income population than in the overall sample, particularly
for women (see Table 4).
Table 4
Prevalence of Having Been Told by a Doctor Have Specific Chronic Conditions
Arthritis
Asthma
Diabetes
High Blood Pressure
Heart Disease
Two or more conditions
General Population
Men
Women
20%
28%
10%
14%
7%
5%
26%
26%
9%
8%
19%
22%
<200% FPL
Men Women
21% 32%
10% 15%
9% 10%
26% 31%
10%
11%
19% 28%
The proportion of low- income adults who reported delaying or not getting a
prescription filled for medication during the previous year by food security status is
shown in Figure 1, within each category of chronic disease and for those with no reported
chronic condition. The relationship of food insecurity, particularly with hunger, with this
index of disease management is striking and consistent across all disease categories, as
well as for those without these illnesses.
12
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Figure 1
Low-Income Adults With a Chronic Disease Who Delayed or Did Not Get
Prescribed Medicine in the Last Year
45
FOOD SECURE
40
FOOD INSECURE W/O HUNGER
35
FOOD INSECURE WITH HUNGER
30
25
20
15
10
5
0
High BP
Heart Disease
Diabetes /
Diabetes / Pills
Asthma
Arthritis
No Disease
Insulin
When delaying or failing to get a prescription filled is examined only in
relationship to drugs or devices prescribed specifically for that condition, the relationship
remains and is consistent, and is particularly striking for individuals with diabetes (Figure
2).
Figure 2
Low-Income Adults With a Chronic Disease Who Delayed or Did
Not Get Prescribed Medicine for that Disease-In the Past Year
30
25
FOOD SECURE
FOOD INSECURE W/O HUNGER
percent
20
FOOD INSECURE WITH HUNGER
15
10
5
0
High BP
Heart
Disease
Diabetes /
Insulin
Diabetes /
Pills
Asthma
Arthritis
13
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
We decided to construct a variable that included not only delay/failure to get
prescriptions filled, but delay and failure to get other recommended types of medical care.
Questions were asked about delay or failure to follow up on getting medical tests or
treatments, and medical referrals to specialty care and to non- medical providers that
would include physical therapists, dietitians, and visiting nurses. Figure 3 shows a
similar presentation for the “delay or did not get care” variable by chronic disease status.
Similar to the prescription variable, food security status was significantly associated with
this failure to get care variable, particularly at the level of food insecurity with hunger.
We therefore used the “delay or did not get care” variable as the dependent variable in the
multivariate analyses.
Figure 3
40
35
30
Low-Income Adults with a Diagnosed Chronic Disease Who
Report Forgoing or Delaying Needed Care* During the Last
Year, by Food Security
FOOD SECURE
FOOD INSECURE W/O HUNGER
percent
FOOD INSECURE WITH HUNGER
25
20
15
10
5
0
High BP
Heart
Disease
Diabetes /
Insulin
Diabetes /
Pills
Asthma
Arthritis
*Prescription Medication, Test or Treatment, or Other Medical Care
Evidence for adverse health outcomes
The only variable in CHIS that gives a reasonably direct measure of adverse
health outcomes in chronic disease is a report of whether the individual visited an
emergency room for care within the last year specifically for complications of their
chronic disease. Figure 4 shows the prevalence of having used an emergency room for
complications of the specific condition by food security status; again, there is a consistent
relationship across all chronic diseases examined. Since earlier analysis showed that
food insecure individuals were not less likely to have a “medical home” than those in
14
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
food-secure low-income households, we cannot attribute this finding to a higher
proportion of individuals using emergency care as primary or only medical care among
the food insecure. The association with food insecurity, particularly at the level of
hunger, is most striking for those conditions for which poor management is likely to
result in acute complications (diabetes, asthma, and heart disease).
Figure 4
Low-Income Adults With a Chronic Disease Who Visited the
Emergency Room For That Disease- In the Past Year
30
percent
25
20
15
10
5
0
High BP
Heart Disease
Diabetes / Insulin
FOOD SECURE
Diabetes / Pills
Asthma
Arthritis
FOOD INSECURE W/O HUNGER
FOOD INSECURE WITH HUNGER
Multivariate predictors of delaying or failing to get needed care for a chronic
disease
The tables in Appendix II–iii through II- vi summarize multivariate logistic
regression models predicting the variable “delaying or failing to get needed care for a
chronic disease in the last year,” for each of the chronic conditions examined. All models
included (i.e., were controlled for) age, gender, ethnicity, household income expressed as
0-99% FPL vs 100-199% FPL, current health insurance (yes or no), family type, and food
security status. Food insecurity, and food insecurity with hunge r, are strong and
consistent independent predictors of failure to get/postponing care; these variables are
significant in all models, with odds ratios of approximately 2 to 5 compared to food
secure individuals. Health insurance, age (middle aged adults at greatest risk) , and
family type also appear in the model for high blood pressure; income and age (middleaged adults at greatest risk) for diabetes; ethnicity is significant only for arthritis, with
American Indian/Alaska Native adults significantly mo re likely to forego or delay care.
Interaction terms (food security status * health insurance, and food security status *
income) did not add significantly to any of the models.
15
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
We moved health insurance, income and food security status variables in and out
of the models (Appendix II- vii through II-xi); the coefficients and odds ratios for all of
these remained remarkably stable.
Puzzled by the less consistent role of health insurance compared to food security
status in these models, we pursued another series of models in which the dependent
variable was constructed as “delaying or failing to get needed for a chronic disease in the
last year because couldn’t afford it,” incorporating the information from the question as
to why a prescription was not filled or other recommended care sought. Although this
strategy reduced the numbers of available cases for analysis (below the viability of the
model for one group, diabetics on insulin), it did shed light on the relative role of health
insurance. The tables in Appendix II, xii- xv show these models. Lack of health
insurance is a strong and consistent predictor of not following through with needed care
because of not being able to afford it, with odds ratios ranging from 1.5 for high blood
pressure to mo re than 10 for heart disease. However, food security status remains
significant in all models, indicating an independent effect, with odds ratios remaining
fairly close to the earlier models at 2 to 5 times higher risk for food- insecure individuals
failing to get or delaying needed care.
Consideration of food assistance. The only food assistance program with
sufficient numbers to consider in these analyses was food stamp program participation.
The numbers of pregnant women, and even of adults with children under five years who
could be assumed to have someone in the family WIC-eligible and had one of the
examined chronic diseases, were relatively small owing to the association of many of
these chronic diseases with age among adults. We will in the future pursue appropriate
analyses for these small groups. In the meantime, however, we selected from the
database adults with household incomes <130% FPL (the food stamp eligibility cutoff)
and re-ran all the models for delaying or failing to get needed care (both in general and
“because couldn’t afford it”) with a food stamp participation variable in the model. The
food stamp participation variable was non-significant in all models.
Discussion
Contrary to our original hypothesis, there was not a striking association of food
insecurity with lack of use of effective preventive medical services, with the exception of
a predictable higher likelihood of having gone more than five years since the last dental
care and less likelihood of having had colorectal and prostate cancer screening tests,
compared to low-income adults in food secure households. Having had a flu shot in the
last year, having participated in effective screening for breast and cervical cancer for
women were equally common across food security categories, and ever having had a
blood cholesterol check was greater than 85% for all. Neither were adults in foodinsecure households less likely to report having a “medical home” than those in foodsecure households. However, adults in food insecure households were significantly
greater users of medical care, measured as number of doctor visits in the previous year
16
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
and history of having used an emergency room in the previous year, as well as likelihood
of having visited a mental health care provider.
Food insecurity was a strong, consistent and independent predictor of failing to
get or delay in getting prescriptions filled and following up on recommended medical
tests and medical and other-provider referrals, even when controlled for household
income (within this low-income sample) and for health insurance. When predictions
were restricted to failure to get or postponement of needed care because of inability to
afford it, food insecurity remained significant in all models, second to and independent of
health insurance in its effect. These finds were true across all the disease categories
examined (arthritis, asthma, heart disease, high blood pressure, diabetes). All of these are
conditions whose effective management requires ongoing medical supervision and
pharmacologic management, and some (diabetes, heart disease) require dietary
management as well. However, they vary widely in the risk of acute or life-threatening
complications when management fails, and likely in the increased medical care costs
associated with poor self- management.
One might ask the question as to whether reverse causality may be operating here;
i.e., does being chronically ill predispose one to food insecurity, other things being equal?
Our earlier analyses of the predictors of food insecurity, while based on cross-sectional
data with all the limitations implied by that constraint, suggest that this is probably not
the case. Prevalence of food insecurity is highest among the lowest- income households,
among Latina, American Indian/Alaska Native and African American households, among
households consisting of a single adult and children; but not in households in which the
index adult reported having one or more of the examined chronic conditions or other
indices of ill health.
There are several aspects of food insecurity in the North American context that
are relevant to interpreting these results as a coherent whole. First, food insecurity in this
environment is most frequent an intermittent, often recurring phenomenon rather than a
consistently persistent condition. This is particularly true for individuals in households
that can be contacted in a telephone-based survey. Thus it may not be surprising that
food insecurity does not appear to interfere with established, one-time or infrequent
behaviors that are health-protective – including having a primary care home, getting a flu
shot, and obtaining mammograms and Pap smears. These results set the context for
arguing that the effects seen in relation to increased medical care use and poorer disease
management (indexed by delay or failure to follow through on recommended care
including prescription drugs, and on increased likelihood of emergency room care for the
specific chronic disease) are not due to more disorganized medical care for individuals
who classify as food- insecure, but rather from personal management decisions made day
by day in the short run and in the face of severe resource constraint. Postponing a
prescription, and either interrupting or diluting the dose, has less immediate
consequences than not feeding one’s children a meal – the consequences may be quite
real and serious a little bit later, but harder for the individual to foresee.
17
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Secondly, it appears from these analyses that the food insecurity variable is
indexing a very fundamental aspect of economic stress, perhaps more generalizable
across domains of resource-requiring needs than other resources. Health insurance is a
powerful predictor of follow-up and management of chronic disease; but even when
health insurance is present, if money is so tight that food security is threatened, healthprotective actions that require ongoing expenditures suffer. The obvious policy
conclusion is that if food security is not present, other resources (such as health insurance)
are less effectively used.
Another aspect of the measurement of food insecurity is that it identifies a
condition characterized most often by both economic constraint and a failure of
management at some level. Poorer mental and emotional health has been repeatedly
identified among adults, adolescents and young children in food- insecure households.
One might argue that poor mental health in adults might predispose to food insecurity,
and indeed in the CHIS data we find that food insecure adults score worse on all of the
mental health variables in the data set than do food-secure low- income adults (data not
shown). However, the findings in other data sets of poorer mental and emotional health
in children and adolescents in food-insecure households can less logically be attributed to
other than the anxiety, uncertainty and chaotic environment that derives from economic
stress severe enough to threaten to compromise food security. Further, investigation into
the characteristics of the 20% of food insecure US households that are not low- income
has indicated that the greatest part of these are characterized by uneven income, changes
in household composition over the previous year, or more than one economic unit within
the household (Nord and Brent, 2002).
The fact that ethnicity does not figure prominently in the models predicting
delaying or not getting needed medical care (only a slight advantage for AfricanAmericans for heart disease and a disadvantage for American Indian/Ala ska Natives for
arthritis were demonstrated) is of interest, since ethnicity is clearly correlated with risk of
household food insecurity in the first place. This finding would argue for an
interpretation that in general, food insecurity operates similarly across cultural groups in
relation to the particular behaviors of interest in this analysis.
Finally, it is appropriate to comment on the lack of influence of food stamp
participation on the delay-or-failure-to-get-care variables. Food stamp coverage is
relatively low in this population (fewer than 20% of CHIS adult respondents who
reported food insecurity with hunger in 2001 participated in the food stamp program). It
has been repeatedly shown that food stamp recipients are more likely to be food insecure
than non-recipients; the present analysis does not add to that observation but only notes
that the relationships between food insecurity and the health care behaviors examined
here are not different when food insecurity co-exists with food stamp participation.
18
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
References Cited
Alaimo K, C Olson, E Frongillo, R Briefel. Food insufficiency, family income and
health in US preschool and school-aged children. Am J Public Health 91: 781-786.
2001b.
Alaimo K, CM Olson, EA Frongillo Jr. Family food insufficiency, but not low family
income, is positively associated with dysthymia and suicidal symptoms in adolescents. J
Nutrition 132: 719-725, 2002.
Alaimo K, CM Olson, EA Frongillo Jr. Food insufficiency and American school-aged
children’s cognitive, academic and psycho-social development Pediatrics 108: 44-53,
2001a.
Blumberg SJ, K Bialososky, WT Hamilton, RR Briefel. The effectiveness off a short
form of the household food security scale. Am J Public Health 89: 1231-1234, 1999.
California Health Interview Survey. CHIS 2001 Methodology Series: Report 4—
Response Rates. Los Angeles, UCLA Center for Health Policy Research, 2002.
Cohen B, Ohls J, Andrews M, Ponza M, Moreno L, Zambrowski A, Cohen R. Food
Stamp Participants’ Food Security and Nutrient Availability. 1999. Mathematica Policy
Research, Inc. 1-119. Alexandria, VA Food and Nutrition Service, USDA.
Hamilton W, Cook JT, Thompson WW, Buron LF, Frongillo EA, Olson CM, Wehler CA.
Household Food Security in the United States in 1995. Summary Report of the Food
Security Measurement Project. September 1997, prepared by Abt Associates, Inc.
Harrison GG, A Stormer, DR Herman, DM Winham. Development of a Spanishlanguage version of the US Household Food Security Survey Instrument. J Nutrition
133:1192-1197, 2003.
Kleinman RE, M Murphy, M Little, M Pagaso, CA Wehler, K Regal, M Jellinek.
Relationship between hunger and psychosocial functioning in low- income American
children. J American Acad Child Adoles Psychiatry 37: 163-170, 1998.
Nelson, K., Brown,M.E. Brown, Lurie, N. Hunger in an Adult Patient Population.
Journal of the American Medical Association 279:1211-1214, 1998.
Nelson K, Cunningham W, Andersen R, Harrison G, Gelberg L. Is food insufficiency
associated with health status and health care utilization among adults with diabetes?
Journal of General Internal Medicine 16:1-8, 2001.
Nicholas T, Vozoris, VS Tarasuk. Household food insufficiency is associated with
poorer health. J Nutrition 133: 120-126, 2003.
19
Harrison GG
July 7, 2004
Draft: Not for citation or circulation
Nord M and CP Brent. Food Insecurity in Higher Income Households. ERS E-FAN
Report No. 02016, 50 pp, 2003. http://www.ers.usda/gov/publications/efan02016/,
accessed July 2004.
Nord M, Jemison K, Bickel G. Prevalence of Food Insecurity and Hunger by State, 19961998. An Economic Research Service Report. ERS/USDA. Sep. 1999
Nord M, M Andrews and S Carlson. Household Food Security in the United States, 2002.
FANRR Report 335, 58 pages. Washington DC: USDA/ERS, 2002.
Rose D, and Oliveira V. Nutrient intakes of individuals from food- insufficient
households in the United States. Am J Public Health 87: 1956-1961, 1997.
Rose, D. Economic Determinants and Dietary Consequences of Food Insecurity in the
United States. Journal of Nutrition 129, Supplement 2s: 517s-520s, 2000.
Siebert K, Heflin C, Corcoran ME, Williams DR. Food insufficiency and the physical
and mental health of low- income women. Women’s Health 59: 159-177, 2001.
Stormer A and GG Harrison. Does household food insecurity predict social and
cognitive development in US kindergartners? Institute for Research on Poverty,
University of Wisconsin Working Paper # DP 1276-03, 2003. Available at:
http://www.ssc.wisc.edu/irp/
Tarasuk VS. Household food insecurity with hunger is associated with women’s food
intakes, health and household circumstances. J Nutrition 131: 2670-2676, 2001.
20
Over 2.2 Million Low-Income California Adults are Food Insecure;
658,000 Suffer Hunger
Gail G. Harrison, Charles A. DiSogra, George Manalo-LeClair, Jennifer Aguayo, Wei Yen
November 3, 2002
More than 2.24 million low-income adults in
California cannot always afford to put food on
the table and as a result almost one out of three of
these adults, 658,000, experience episodes of
hunger. This is a sad reality in a state that has the
largest agricultural economy in the United States
and produces an abundance of high-quality fruits
and vegetables for much of the nation. The ranks
of “food insecure” Californians include not just
the most impoverished individuals but working
adults, retired older persons with fixed incomes
and many parents with children.
These new findings are based on data from the
California Health Interview Survey (CHIS 2001).
CHIS 2001 is California’s largest representative
health survey of the state and its counties. The
survey included a sizeable sample of the
estimated 8 million low-income Californians –
those living in households with incomes below
200% of the federal poverty level* . It was found
that more than 8.3% of these low- income adults
experience food insecurity with hunger. Another
20.0%, one out of five low- income adults,
experience food insecurity that falls short of
hunger. Food insecurity, with or without hunger,
causes families to forego such basic needs as
rent, utilities, and medical care in order to put
food on the table. Food security is defined as
access to an adequate nutritious diet. Food
security is a goal of any society; essential for the
good health of all. Based on these new CHIS
2001 findings, the paradox of food insecurity and
hunger in food-abundant California clearly shows
that this state can do better.
*
Example: 200% FPL is $36,200 per year for a family of
four.
11/03/02
Measuring Food Insecurity
Lack of assured access to enough food through
socially acceptable means is termed “food
insecurity.” In its extreme form, this results in
hunger - going without food for lack of money or
other resources. Over the last several decades,
health advocates and researchers have worked on
ways to accurately measure the prevalence of
hunger and food insecurity in order to track
trends using this basic indicator of human
welfare. These efforts have resulted in the
development of standard instruments to estimate
the prevalence and severity of the problem. The
food security measure that was used in CHIS
2001 is an abbreviated six- item scale derived
from the 18- item US Household Food Security
Module used in national surveys 1 . In CHIS 2001,
the food security questions were asked only of
individuals in households whose incomes were
estimated to be less than 200% of the federal
poverty level. The survey asked about a person’s
food security over the previous 12-month period
and focused on the lack of resources or money as
reasons for food insecurity. The survey only
interviewed persons living in households with
telephones. Households without telephones and
the homeless population are not included in these
results.
Food Insecurity is a Risk to Health
While closely associated with poverty, food
insecurity is a threat to well-being and long-term
health. There is abundant evidence from other
studies that hunger and food insecurity pose
substantial risks to health resulting in large costs
to society through increased needs for medical
care. There are also related social and mental
health costs. Individuals who are food insecure
have been shown to have poor quality diets
making them vulnerable to a wide variety of
diseases.
There are also health risks and
UCLA Center for Health Policy Research
Page 1 of 10
consequences related to the worry, anxiety, and
management tradeoffs that must be employed in
food- insecure households. Whether mediated
through malnutrition or other routes, the negative
impacts are substantial. Children living in foodinsecure households tend to do less well in school
with increased absences, tardiness, more school
suspensions, and poorer cognitive functioning2-4 .
Their overall health status is worse with more
health problems such as headaches, colds, and
ear infectio ns than other children5,6. Adolescents
in food- insecure households have higher rates of
depressive and suicidal symptoms and are more
than twice as likely to have seen a psychologist
than other teenagers7 . Adults who are food
insecure tend to forego needed medical care
when hunger threatens. For example, food
insecure adults with diabetes have more costly
and life-threatening complications of their
illness; requiring more physician visits and
medical care than food-secure persons with
diabetes8 .
Food Insecurity and Hunger in California
CHIS data show that 28.3% of low-income adults
in California are food insecure; this represents
about 2.24 million adults. Although this study
only looked at low- income adults, it should be
understood that some fraction of adults above
200% of poverty also experience food insecurity
intermittently, seasonally or in response to
sudden shocks such as job layoffs or illness.
Thus, it is reasonable to conclude that among all
California adults, the number of food insecure is
higher than 2.24 million. When children in these
households are taken into account, the actual
number of persons may exceed 5 million
according to estimates by the U.S. Department of
Agriculture9 .
Based on the recent CHIS 2001 data, among the
poorest adults, those in households with incomes
below 100% of the federal poverty line, 36.2% or
1.2 million adults had experienced food
insecurity sometime in the twelve months prior to
the survey (see Exhibit 1).
Among adults living in households with incomes
between 100% and 199% of poverty, 22.7% or
almost 1.05 million were food insecure. Among
all adults in households with incomes below
200% of poverty, 28.3%, or 2.24 million were
food insecure in 2001. The CHIS 2001 results
show that for all adults below 200% of poverty
who are food insecure, one in three experience
hunger. Statewide this is estimated to be 658,000
persons. More than half of these persons, 365,000
are below 100% of poverty.
The level of food insecurity and hunger among
low- income adults varies across the state.
Exhibit 2 is a map showing the geographic
distribution of food insecurity across California
in the 33 counties and 8 county groupings
measured in the survey* . Many of California’s
northern rural counties and the agriculturally rich
San Joaquin Valley have the highest rates of food
insecurity, exceeding 30% of low- income
households in several northern counties and
ranging from 33% to 41% in the San Joaquin
Valley. The highest rates are found in Tulare
(41.4%) and Fresno (35.7%) counties. Exhibit 3
shows the percent of low- income adults who are
food insecure and the percent that are food
insecure with and without experiencing hunger
for each of the 41 counties and geographic areas
surveyed. Hunger was also prevalent among
low- income adults across most of the state’s
northern rural Sierra range counties and across
the San Joaquin Valley. The highest rates of
hunger were reported for the Humboldt-Del
Norte area (15.9%) and Shasta County (15.6%).
In the more populous urban areas, Los Angeles
County has a food insecurity prevalence of
30.1% among low- income households and, due
to its large population, contributes the largest
number of the state’s adults, approximately
777,000 living under the threat of hunger or
actually experiencing hunger. Among the San
Francisco Bay Area’s low- income adults, food
insecurity ranges from 22% to 31% across
*CHIS sampled 33 individual counties and 8 geographic
areas made up of groupings of smaller counties. When
combined, these 41 counties and geographic areas
encompass the entire state.
11/03/02
UCLA Center for Health Policy Research
Page 2 of 10
counties.
The highest rates of hunger are
reported in Sonoma, Solano, Marin and Napa
counties.
Among race/ethnic groups, the highest rates of
food insecurity are among low- income American
Indian/Alaska Natives, African Americans, and
Latinos. American Indian/Alaska Natives and
African Americans report the most severe
problems reflected in the highest hunger rates of
16.9% and 14.6%, respectively (see Exhibit 4).
Among the most vulnerable adults in the
population in terms of the effects of being
hungry, are pregnant women and the elderly.
Approximately 14.4% of older, low- income
adults, those over 65 years old, were found to be
food insecure. Among low- income pregnant
women, the prevalence of food insecurity was
29.5%. As many as 41.9% of low- income adults
in single-parent families with children were food
insecure and one out of three of these
experienced hunger.
Among low- income
families, being unemployed and looking for work
conferred additional risk. There was a 39.1%
food insecurity prevalence among these persons
contrasted with 27.8% for those who reported
working. Undocumented individuals seem to be
at particular risk; 35.8% of low-income adults
who reported not being legal residents were foodinsecure compared to 25.2% of US citizens of
similar income levels (see Exhibit 5).
Food Assistance Programs
Good jobs with livable wages and an adequate
social safety net for people who are unable to
work are California’s primary defenses against
hunger. Absent these, the next best response to
hunger and food insecurity is wide participation
in the federal food assistance programs. Each
program [see box] has a unique and necessary
role in reduc ing food insecurity.
The Food Stamp Program: Enables lowincome children, families and individuals to buy
nutritious food through normal channels like
supermarkets and grocers.
Women, Infants and Children Special
Supplemental Nutrition Program (WIC): This
highly effective program sees that low- income
pregnant women and their babies, and young
children determined to be at nutritional risk have
access to the nutritious food and prenatal care
they need.
The CHIS 2001 data show that some adults in
need of help are participating in the federal Food
Stamp program. Also, women in most need are
likely to be enrolled in the Women, Infants and
Children Special Supplemental Nutrition
Program (WIC). Among the estimated 4.95
million adults in families with incomes low
enough to be eligible for food stamps, that is,
below 130% of the federal poverty level, about
496,000 or 10.2% reported receiving food
stamps. Of these Food Stamp participants,
51.1% had experienced food insecurity in the
previous year compared to only 27.7% of the
non-participants. However, among adults below
130% of poverty who experience hunger, 80.5%
(approximately 358,000 adults) were not in the
Food Stamp program.
For all pregnant women who are below WIC’s
income eligibility criterion, that is, 185% of the
poverty level, 31.7% or almost 31,000 report
being food insecure. One out of four of these
food insecure women (24.1%) report not being
enrolled in WIC. Among these 31,000 food
insecure pregnant women, over 8,300 reported
experiencing hunger with one in three (34.2%)
not being enrolled in the WIC program (see
Exhibit 6).
Nutritional risk as well as income are criteria
used for WIC eligibility. Just looking at the
estimated 1 million income-eligible women who
reported being pregnant and/or had one or more
children under age 5, almost 58.5% reported
being enrolled in WIC with two out of five of
these women being food insecure. And although
11/03/02
UCLA Center for Health Policy Research
Page 3 of 10
about 8% or 78,000 of this population of women
reported hunger whether or not they were
enrolled in WIC, the WIC enrollees tended to be
poorer (70.7% below the poverty line vs. 46.2%
for non-enrollees), younger and much more
likely to be unemployed. Since poverty is
associated with hunger, these data suggest that
the WIC program is enrolling those in greatest
need. This also suggests that WIC is potentially
checking hunger among participants because the
hunger rate is not greater than that of nonparticipants.
While these federal programs help reduce hunger
and food insecurity, they are underutilized. The
Food Stamp Program reaches only 49% of
eligible Californians according to the U.S.
Department of Agriculture10 . There are several
determining factors for eligibility for food
stamps, including income. Households may be
eligible for food stamps if their incomes are
below 130% of the federal poverty level. CHIS
suggests that 1.46 million adults experiencing
food insecurity in California have incomes below
this 130% level, yet almost 1.21 million are not
getting food stamps. The CHIS findings indicate
that this is resulting in unnecessary hunger
because food stamps can and do help.
Policy Recommendations
It is no surprise that poverty and hunger go hand
in hand.
Increases in the resources that
Californians have for food can be achieved
through improved wages and adequate cash
assistance programs for seniors, the disabled and
the unemployed and underemployed. In the
meantime, the following policy options should be
considered.
1. Hunger and food insecurity should be
routinely included as basic health indicators
in all health surveillance surveys in
California, as it is in national surveys.
2. Increase participation in the federal food
programs by streamlining enrollment
procedures. State policymakers can make
application processes easier, especially for
11/03/02
busy working
following:
families,
by
doing
the
a. Reduce trips to the food stamp office.
It takes an average of 5 hours and 3
trips to a county office to get food
stamps11 . Legislators can require the
use of appropriate alternatives, such
as telephone and mail- in applications.
b. Automatically enroll low- income
children in school meals. California is
just one of three states that does not
match state welfare data with state
education data to automatically enroll
its children in the free and reduced
priced school meal programs.
c. Eliminate the practice of fingerimaging applicants.
Food Stamp
applicants in California must submit
to a finger print type scan in order to
get help. This costly imaging adds
time to an already lengthy process.
The California State Auditor is
currently examining if and how this
may be deterring eligible people from
applying.
3. Seize the opportunity provided by the
Congressional Reauthorization of Child
Nutrition Programs in 2004. As the data
indicate, families with children, especially
single-parent families, had higher rates of
food insecurity. Congress can address this
next year when it reviews child nutrition.
Among the things Congress can do:
a. Expand the reach of the School
Breakfast Program and move toward a
“universal” breakfast system, where
all students can start the day with
breakfast regardless of income. It can
also promote the “breakfast in the
classroom” alternative that is intended
to remove the problems of space and
time for breakfast at many schools.
b. Ensure children are fed year round by
making it easier for community-based
organizations to serve meals when
UCLA Center for Health Policy Research
Page 4 of 10
school is out by combining the
Summer Food Program and Child
Care Food program.
c. Serve more children in school by
offering all low-income children free
meals. Schools offer meals in three
categories: full-price, reduced-price
and free. The reduced price category
serves children between 130% and
185% of poverty and these students
are required to partially pay for their
meal. This partial payment can be
burdensome for many low-wage
families and is also burdensome for
schools to administer.
4. Invest in outreach for the Food Stamp
program and WIC since their target
populations are the groups with the highest
prevalence of food insecurity and hunger.
***
Author Information
Gail G. Harrison is Professor in the Department of
Community Health Sciences at the School of Public
Health at UCLA, and Associate Director of the
UCLA Center for Human Nutrition, Los Angeles.
George Manalo-LeClair is Director of Legislation and
Policy for California Food Policy Advocates. Charles
A. DiSogra is Director of the California Health
Interview Survey at the UCLA Center for Health
Policy Research with Jennifer Aguayo, CHIS
Research Associate, and Wei Yen, CHIS Associate
Director for Research.
Acknowledgements
The authors appreciate the contributions of:
Hongjian Yu, Lu -May Chiang, Lida Becerra, Jenny
Chia, Rong Huang, Carolyn Mendez and reviewers E.
Richard Brown and Roberta Wyn.
References
1.
2.
Data Source
This policy brief on food insecurity and hunger in
California is based on findings from the 2001
California Health Interview Survey (CHIS 2001).
CHIS 2001, the largest health survey ever conducted
in any state and one of the largest in the nation,
covers a broad range of public health concerns
including health status and conditions, health-related
behaviors, health insurance coverage and access to
health care services. CHIS 2001 interviewed 55,428
households randomly drawn from every county in
California for its random-digit dial (RDD) telephone
survey, providing a sample that is representative of
the state’s non-institutionalized population (data were
weighted based on the 2000 Census). CHIS 2001
interviewed one sample adult in each household. The
data on food insecurity and hunger are derived from
the adult interviews from households with incomes
below 200% of the federal poverty definition. The
interviews, available in six languages, were
conducted between November 2000 and September
2001.
11/03/02
3.
4.
5.
6.
7.
Blumberg SJ, Bialososky K, Hamilton WL,
Briefel RR (1999). The effectiveness of a short
form of the household food security scale. Am J
Public Health 89: 1231-1234.
Alaimo K, Olson CM, Frongillo EA Jr. Food
insufficiency and American school-aged
children’s cognitive, academic, and psychosocial
development. Pediatrics 108: 44-53, 2001.
Kleinman RE, Murphy JM, Little M, Pagano M,
Wehler, CA, Regal K, Jellinek MS. Hunger in
children in the United States: Potential behavioral
and emotional correlates. Pediatrics 101: 1-6,
1998.
Murphy JM, Wehler CA, Pgano ME, Little M,
Kleinman RE, Jellinek MS. Relationsship
between hunger and psychosocial functioning in
low-income American children. J American
Academy of Child & Adolescent Psychiatry 37:
163-170, 1998.
Alaimo K, Olson CM, Frongillo EA Jr, Briefel
RA. Food insufficiency, family income and
health in US preschool and school-aged children.
Am J Public Health 91: 781-786, 2001.
Casey PH, Szeto K, Lensing S, Bogle M, Weber J.
Children in food-insufficient, low-income
families: prevalence, health and nutritional status.
Archives of Pediatrics and Adolescent Medicine
155: 508-514, 2001.
Alaimo K, Olson CM, Frongillo EA Jr. Family
food insufficiency, but not low family income, is
positively associated with dysthymia and suicide
symptoms in adolescents. J Nutrition 132; 719725, 2002.
UCLA Center for Health Policy Research
Page 5 of 10
8.
Nelson K, Cummingham W Anderson R. Harrison
G. Gelberg L Does food insufficiency affect
health status and health care utilization among
diabetics? Data from NHANES III. Am. J.
Internal Medicine 16:404-411, 2001.
Sullivan AF, Choi E (2002). Hunger and Food
Insecurity in the Fifty States, 1998-2000. Boston:
Brandeis University Center on Hunger and
Poverty.
9.
10. Reaching those in Need: State Food Stamp
Participation Rates, June 2002. Mathematica
Policy Research for USDA.
http://www.fns.usda.gov/oane/MENU/Published/F
SP/FILES/Participation/1999rates.pdf
11. Ponza M, Ohls JC, Moreno L, Zambrowski A,
Cohen R. (1999). Customer Service in the Food
Stamp Program. Mathematica Policy Research,
Inc. for USDA.
http://www.fns.usda.gov/oane/MENU/Publish
ed/FSP/FILES/Program%20Operations/fspcu
st.pdf
Exhibit 1.
Number of Low-Income Adults Reporting Food Insecurity and Hunger in California
Food Insecure with Hunger
Food Insecure without Hunger
2.243
2.25
millions of adults
2.00
1.75
1.50
1.25
1.585
1.196
1.047
1.00
0.75
0.831
0.754
0.50
0.25
0.658
0.365
0.293
<100%
100%-199%
0.00
All <200%
percent of poverty level
Source: 2001 California Health Interview Survey
11/03/02
UCLA Center for Health Policy Research
Page 6 of 10
Exhibit 2.
11/03/02
UCLA Center for Health Policy Research
Page 7 of 10
Exhibit 3.
Prevalence of Food Insecurity (with and without hunger) among Adults (Ages 18+)
below 200% poverty, by County/County Group*,
California, 2001
Food Insecure
NORTHERN AND SIERRA COUNTIES
Butte
Shasta
Humboldt, Del Norte
Siskiyou, Lassen, Trinity, Modoc
Mendocino, Lake
Tehama, Glenn, Colusa
Sutter, Yuba
Nevada, Plumas, Sierra
Tuolumne, Calaveras, Amador, Inyo,
Mariposa, Mono, Alpine
GREATER BAY AREA
Santa Clara
Alameda
Contra Costa
San Francisco
San Mateo
Sonoma
Solano
Marin
Napa
SACRAMENTO AREA
Sacramento
Placer
Yolo
El Dorado
SAN JOAQUIN VALLEY
Fresno
Kern
San Joaquin
Stanislaus
Tulare
Merced
Kings
Madera
CENTRAL COAST
Ventura
Santa Barbara
Santa Cruz
San Luis Obispo
Monterey, San Benito
LOS ANGELES
Los Angeles
OTHER SOUTHERN CALIFORNIA
Orange
San Diego
San Bernandino
Riverside
Imperial
STATEWIDE
Prevalence 95% C.I.**
24.2
(18.8-29.5)
35.0
(28.1-41.8)
31.8
(25.6-38.0)
30.8
(24.6-37.0)
28.2
(21.9-34.4)
32.6
(26.7-38.6)
31.8
(25.5-38.0)
19.8
(11.9-27.7)
Food Insecure
WITHOUT Hunger
Prevalence
95% C.I.
12.9
(8.9-17.0)
19.4
(13.3-25.4)
16.0
(11.2-20.7)
18.9
(13.7-24.1)
16.2
(10.5-21.8)
20.1
(15.1-25.2)
21.1
(15.4-26.8)
15.9
(8.1-23.6)
Food Insecure
WITH Hunger
Prevalence
95% C.I.
11.2
(7.2-15.2)
15.6
(10.5-20.7)
15.9
(10.9-20.8)
11.9
(7.3-16.5)
12.0
(8.0-16.0)
12.5
(8.1-16.9)
10.7
(6.7-14.6)
4.0
(1.4-6.6)
25.2
(18.5-31.7)
14.8
(9.4-37.0)
10.3
(5.6-15.1)
25.1
21.7
21.6
28.8
22.0
28.9
30.2
26.1
31.4
(18.4-31.8)
(16.4-27.0)
(14.5-28.7)
(23.7-33.9)
(14.0-30.1)
(19.7-38.1)
(23.8-36.6)
(14.5-37.7)
(22.1-40.6)
19.9
14.7
17.1
19.4
15.5
14.8
18.8
15.3
20.1
(13.5-26.3)
(10.1-19.3)
(10.3-23.9)
(14.8-24.1)
(8.4-22.6)
(7.4-22.1)
(12.9-24.7)
(6.6-23.9)
(12.1-28.1)
5.2
7.0
4.5
9.4
6.5
14.1
11.4
10.8
11.3
(2.3-8.1)
(4.1-10.0)
(1.8-7.1)
(6.6-12.2)
(2.1-10.9)
(7.2-21.1)
(7.7-15.1)
(1.2-20.4)
(4.6-18.0)
24.0
19.0
22.9
22.0
(18.4-29.7)
(10.9-27.0)
(15.8-29.9)
(15.1-28.9)
15.9
11.4
14.3
9.0
(10.9-21.0)
(5.3-17.4)
(8.1-20.5)
(4.2-13.7)
8.1
7.6
8.6
13.0
(4.8-11.4)
(1.5-13.7)
(4.2-12.9)
(7.5-18.5)
35.7
33.6
32.6
33.0
41.4
34.1
34.6
32.9
(29.9-41.4)
(28.6-38.7)
(26.7-38.5)
(25.9-40.1)
(34.7-48.0)
(28.0-40.3)
(27.9-41.4)
(27.0-38.8)
24.6
25.4
21.0
22.6
30.8
23.7
25.9
21.3
(19.5-29.7)
(20.7-30.0)
(15.8-26.1)
(16.3-28.9)
(24.6-37.0)
(18.0-29.5)
(19.4-32.5)
(16.2-26.4)
11.1
8.3
11.7
10.4
10.6
10.4
8.7
11.6
(7.0-15.2)
(5.3-11.2)
(7.9-15.5)
(5.8-15.0)
(6.5-14.6)
(6.8-14.0)
(5.3-12.1)
(7.6-15.6)
19.8
22.8
24.5
25.5
29.4
(13.2-26.4)
(17.4-28.1)
(17.6-31.4)
(19.0-32.0)
(22.1-36.7)
15.1
16.3
16.2
13.0
21.9
(9.4-20.7)
(11.6-21.0)
(10.1-22.3)
(8.2-17.8)
(15.0-28.7)
4.8
6.5
8.3
12.5
7.5
(0.7-8.8)
(3.6-9.4)
(4.2-12.4)
(7.4-17.7)
(3.9-11.1)
30.1
(28.4-31.8)
22.0
(20.4-23.5)
8.1
(7.2-9.0)
25.5
26.9
28.5
27.0
26.2
28.0
(21.2-29.8)
(23.1-30.8)
(24.0-32.9)
(22.3-31.7)
(21.3-31.1)
(27.4-29.3)
19.6
19.0
19.2
18.1
21.3
20.0
(15.5-23.7)
(15.4-22.5)
(15.3-23.0)
(13.9-22.1)
(16.7-25.9)
(19.2- 20.9)
5.9
8.0
9.3
9.0
4.9
8.0
(3.9-7.8)
(5.8-10.1)
(6.5-12.1)
(6.0-11.9)
(2.7-7.1)
(7.8- 8.8)
Source: 2001 California Health Interview Survey
*CHIS sampled 33 individual counties and 8 geographic areas made up of groupings of smaller counties. When combined, these 41
counties and geographic areas encompass the entire state.
** The prevalence results represent estimated values that are very close to the actual values for adults (ages18+) living below 200% of
poverty who experienced food insecurity (with and without hunger) in California in 2001. Because the estimated value is based on a
sample of this population, it has a degree of uncertainty, and the confidence interval (C.I.) shows the range where the actual value may lie.
Hence, for 95% C.I., you can assume with 95% confidence that the actual value lies between the lower and upper C.I. range.
11/03/02
UCLA Center for Health Policy Research
Page 8 of 10
Exhibit 4.
Prevalence of Food Insecurity (with and without hunger)
and Prevalence of Hunger by Race/Ethnicity*
for Adults (Ages 18+), Below 200% Poverty
=Food insecure
(with and without hunger)
=Hunger
California , 2001
50%
45%
40%
37.2
35%
35.2
32.7
30%
26.6
27.9
25%
25.3
22.4
20%
15%
16.9
14.6
13.9
10%
9.7
9.1
7.5
5%
4.0
0%
White
Latino
Asian
African American
AIAN
NHOPI
Others
* Race/Ethnicity is presented here using the UCLA Center for Health Policy Research's definition which treats Latino as a mutually exclusive category and tak es into account the
race/ethnicity with which respondents most identify.
**The bold, vertical bars represent the 95% confidence interval (C.I.) bands. Because the estimated value is based on a sample of this population, it has a degree of uncertainty,
and the C.I. bands shows the range where the actual value may lie.
Note: American Indian and Alaska Native is abbreviated AIAN. Native Hawaiian and Other Pacific Islander is abbreviated NHOPI.
Source: 2001 California Health Interview Survey
Exhibit 5.
Prevalence of Food Insecurity (with and without hunger)
Among Vulnerable Adult Groups,
Below 200% Poverty,
California, 2001
Food insecure with hunger
45%
Food insecure without hunger
41.9
39.1**
40%
35.8
35%
29.5
30%
28.3
25.1
25%
28.5
20%
14.4*
22.2
15%
10%
10.8
14.1
13.6
5%
7.3
7.3
3.6
0%
Older Adults (Age 65+)
Pregnant Women
(Ages 18-44)
Single Adults
with Children
Unemployed Adults
Looking for Work
Undocumented
Adults
*Top numbers represent total prevalence of food insecurity (with and without hunger) for each group.
**Total prevalence of food insecurity does not add up due to rounding.
Source: 2001 California Health Interview Survey
11/03/02
UCLA Center for Health Policy Research
Page 9 of 10
Exhibit 6.
Percent of Income -Eligible Person Reporting Hunger
Participating in Food Assistance Programs
(FPL= Federal Poverty Level)
Adults Below 130% FPL
Yes
19.5%
Pregnant Women below
185% FPL
No
34.2%
Yes
65.8%
No
80.5%
Food Stamps Participation
WIC Participation
Source: 2001 California Health Interview Survey
For further information contact:
Charles DiSogra, DrPH
Director, California Health Interview Survey
UCLA Center for Health Policy Research
(310) 794-0946 cdisogra@ucla.edu
11/03/02
UCLA Center for Health Policy Research
Page 10 of 10
Appendix II. Multivariate prediction models
i.
ii.
iii.
iv.
v.
vi.
vii.
viii.
ix.
x.
xi.
xii.
xiii.
xiv.
xv.
Predictors of household food insecurity
Predictors of household food insecurity with hunger
Significant predictors of postponing or not getting needed care for high blood pressure
Significant predictors of postponing or not getting needed care for diabetes
Significant predictors of postponing or not getting needed care for arthritis
Significant predictors of postponing or not getting needed care for heart disease and asthma
Income, insurance and food security as predictors of delaying or not getting care for high blood pressure
Income, insurance and food security as predictors of delaying or not getting care for diabetes, among patients taking oral
hypoglycemic agents
Income, insurance and food security as predictors of delaying or not getting care for diabetes, among patients taking insulin
Income, insurance and food security as predictors of delaying or not getting care for asthma
Income, insurance and food security as predictors of delaying or not getting care for arthritis
Predictors of delaying or not getting care for high blood pressure because could not afford it
Predictors of delaying or not getting care for diabetes, amo ng patients taking oral hypoglycemic agents, because could not
afford it
Predictors of delaying or not getting care for arthritis, because could not afford it
Predictors of delaying or not getting care for heart disease and asthma, because could not afford it
II – i. Predictors of Food Insecurity among Low-Income California Adults
Variable
Household Income
<100% FPL
100-130% FPL
131-185% FPL
186-199% FPL
Gender
Male
Female
Age (years)
18-35
36-50
51-64
65+
Ethnicity
White
Latino
African American
AIAN
AAPI/Other
Family Type
Married no children
Married w/ children
Single no children
Single w/ children
Current health insurance
Yes
No
Chronic illnesses
None
One
More than one
Beta
p
Odds Ratio
Referent
-0.84
-0.56
-0.15
<.00001
<.00001
n.s.
Referent
0.02
n.s.
1.02
0.92-1.14
Referent
0.39
0.24
-0.57
<.00001
<.005
<.00001
1.48
1.27
0.57
1.31-1.67
1.09-1.49
0.48-0.68
Referent
0.16
0.29
0.46
0.04
<.05
<.005
<.05
n.s.
1.17
1.33
1.59
1.04
1.03-1.33
1.10-1.61
1.10-2.28
0.87-1.24
Referent
0.35
0.24
0.71
.0001
<.005
<.00001
1.41
1.27
2.03
1.19-1.68
1.09-1.49
1.66-2.49
Referent
0.21
<0005
1.23
1.10-1.39
Referent
-0.23
0.17
<.05
n.s.
0.79
1.19
0.66-0.95
0.99-1.43
0.43
0.57
0.86
Wald Chi-Square for Model: 2060.24, p<.00001; Minus intercept: 629.84, p<.00001
95% Confidence Interval
0.37-0.51
0.51-0.65
0.72-1.02
II-ii. Predictors of Food Insecurity with Hunger among Low-Income California Adults
Variable
Household Income
<100% FPL
100-130% FPL
131-185% FPL
186-199% FPL
Gender
Male
Female
Age (years)
18-35
36-50
51-64
65+
Ethnicity
White
Latino
African American
AIAN
AAPI/Other
Family Type
Married no children
Married w/ children
Single no children
Single w/ children
Current health insurance
Yes
No
Chronic illnesses
None
One
More than one
Beta
p
Odds Ratio
95% Confidence Interval
Referent
-0.88
-0.65
-0.00
<.00001
<.00001
ns
0.41
0.52
1.00
0.33-0.53
0.43-0.63
0.79-1.26
Referent
0.11
ns
1.11
0.95-1.31
Referent
0.67
0.52
-0.78
<.00001
<.00001
<.00001
1.95
1.67
0.46
1.62-2.33
1.33-2.11
0.34-0.62
Referent
-0.50
0.14
0.38
-0.67
<.00001
ns
ns
<.00001
0.60
1.15
1.46
0.51
0.51-0.72
0.89-1.49
0.97-2.21
0.39-0.67
Referent
0.25
0.37
0.58
ns
<.005
.0001
1.28
1.45
1.79
0.96-1.71
1.12-1.88
1.33-2.42
Referent
0.16
ns
1.17
0.99-1.38
Referent
-0.57
-0.11
<.00001
ns
0.56
0.90
0.43-0.73
0.69-1.17
Wald Chi-Square for Model: 4076.98, p<.0000; Minus intercept: 420.79, p<.0001
Table II-iii. Significant Predictors of Postponing or Not Getting Needed Care for High Blood Pressure*
(“Care” refers to filling prescription for medicine, getting recommended
medical tests, and/or following through on referrals to medical or other health care providers)
Coefficient
No health insurance
.58
Significance Odds Ratio
95% Confidence Intervals
<.01
1.8
1.2-2.8
Age 36-50 years 1
1.21
<.0001
3.3
1.9-5.8
Age 50-64 years 1
.76
.01
2.1
1.2-3.9
Single without children2
- .37
<.05
0.7
0.4-.99
Food insecure without hunger3
.54
<.01
1.7
1.2-2.8
Food insecure with hunger3
1.2
<.0001
3.3
2.2-5.0
1
Referent is 18-35 years.
Referent is married with children.
3
Referent is food secure.
2
*Model also includes gender, ethnicity, household income as % of poverty.
Number of cases in the analysis: 5153
Table II-iv. Significant Predictors of Postponing or Not Getting Needed Care for Diabetes*
(“Care” refers to filling prescription for medicine, getting recommended
medical tests, and/or following through on referrals to medical or other health care providers)
Coefficient
Taking insulin
Food insecure without hunger
95% Confidence Intervals
1.17
<.05
3.2
1.01-10.2
1.46
<.005
4.3
1.6-11.9
<.05
4.5
1.2-17.2
Household income <100% poverty2 -.58
.05
0.6
0.3-1.0
Food insecure without hunger
1.14
.001
3.1
1.6-6.2
Food insecure with hunger
1.56
<.00001
4.8
2.5-9.1
Food insecure with hunger
Taking oral drugs
Age 36-50 years 1
1
Significance Odds Ratio
1.50
Referent is 18-35 years.
Referent is 100-199% poverty.
2
*Models also include gender, ethnicity, health insurance, family type.
Number of cases in the models: 401 for diabetics taking insulin, 1072 for diabetics taking oral drugs.
Table II-v. . Significant Predictors of Postponing or Not Getting Needed Care for Arthritis*
(“Care” refers to filling prescription for medicine, getting recommended
medical tests, and/or following through on referrals to medical or other health care providers)
Coefficient
Significance Odds Ratio
95% Confidence Interval
American Indian/Alaska Native1
.89
<.01
Food insecure without hunger2
.70
.0001
2.0
1.4-2.8
<.00001
4.8
3.4-6.7
Food insecure with hunger2
1
2
1.56
2.4
1.3-4.6
Referent is white.
Referent is food secure.
*Model includes age group, gender, household income as % of poverty, family type, health insurance.
Number of cases in the analysis: 4996
Table II-vi. Significant Predictors of Postponing or Not Getting Needed Care for Heart Disease and Asthma*
(“Care” refers to filling prescription for medicine, getting recommended
medical tests, and/or following through on referrals to medical or other health care providers)
Coefficient
Significance Odds Ratio
95% Confidence Interval
Heart Disease
Food insecure with hunger1
.82
<.01
2.3
1.2-4.7
Asthma
Food insecure with hunger1
.60
<.01
1.8
1.2-2.9
1
Referent is food secure.
*Models include gender, age group, ethnicity, household income as % of poverty, family type, health insurance.
Number of cases in the analysis: 1937 for heart disease, 2295 for asthma.
Table II-vii. Income, Insurance and Food Security as Predictors of Delay or Not Getting Care for High Blood Pressure
(Odds Ratios and significance of beta coefficient; models adjusted for gender, age, ethnicity, and family type)
Full Model
OR
p
-Income
OR p
Income <poverty1
0.92
ns
0.95
No health insurance2
1.74
.01
1.73
Food insecure3
no hunger
with hunger
1.71 <.01
3.28 <.00001
1
2
3
Referent is 100-199% poverty level
Referent is any health insurance
Referent is food secure
ns
-Insurance
OR
p
1.02
ns
.01
1.69 <.01
3.24 <.00001
-Food Security
OR
p
1.70
1.73 .005
3.21 <.00001
.01
Table II-viii. Income, Insurance and Food Security as Predictors of Delay or Not Getting Care for Diabetes
Among Patients Taking Oral Hypoglycemic Agents
(Odds ratios and significance of beta coefficients; models adjusted for gender, age, ethnicity and family type)
Full Model
OR
p
Income <poverty1
.57
.05
No health insurance
2.89
ns
Food insecure3
no hunger
with hunger
3.13 .001
4.73 <.00001
1
Referent is 100-199% poverty level
Referent is any health insurance
3
Referent is food secure
2
-Income
OR
p
-Insurance
OR
p
57
.88
.05
ns
2.73 <.005
4.47 <. 00001
-Food Security
OR
p
.64
.78
3.14 <.005
4.75 <.00001
ns
ns
Table II-ix. Income, Insurance and Food Security as Predictors of Delay or Not Getting Care for Diabetes
Among Patients Taking Insulin
(Odds ratios and significance of beta coefficients; models adjusted for gender, age, ethnicity, and family type)
Full Model
OR
p
Income <poverty1
.62
ns
No health insurance2
1.49
ns
Food insecure3
no hunger
with hunger
3.17
4.35
.05
<.005
1
Referent is 100-199% poverty level
Referent is any health insurance
3
Referent is food secure
2
-Income
OR
p
-Insurance
OR
p
.61
1.53
ns
ns
2.84 ns
4.04 <.005
3.22 <.05
4.45 <.005
-Food Security
OR
p
.67
ns
1.81
ns
Table II-x. Income, Insurance and Food Security as Predictors of Delay or Not Getting Care for Asthma
(Odds ratios and significance of beta coefficients; model adjusted for gender, age, ethnicity, family type)
Full Model
OR
p
-Income
OR
p
Income <poverty1
0.97
No health insurance2
1.36
ns
1.36
ns
Food insecure3
no hunger
with hunger
0.77
1.82
ns
<.01
0.77
1.81
ns
<.01
1
Referent is 100-199% poverty level
Referent is any health insurance
3
Referent is food secure
2
ns
-Insurance
OR
p
-Food Security
OR
p
-0.02
-0.01
ns
1.34
ns
ns
0.78 ns
1.82 <.01
Table II-xi. Income, Insurance and Food Security as Predictors of Delay or Not Getting Care for Arthritis
(Odds ratios and significance of beta coefficients; model adjusted for gender, age, ethnicity, family type)
Full Model
OR
p
Income <poverty1
0.86 ns
No health insurance2
1.39
Food insecure3
no hunger
with hunger
2.02 .0001
4.81 <.0001
1
Referent is 100-199% poverty level
Referent is any health insurance
3
Referent is food secure
2
ns
-Income
OR
p
-Insurance
OR
p
0.86
1.38
ns
ns
1.99 .0001
4.68 <.0001
2.03 .0001
4.79 <.0001
-Food Security
OR
p
0.98
ns
1.36
ns
Table II-xii. Predictors of Delaying or Not Getting Care for High Blood Pressure because Could Not Afford It
(Model adjusted for gender, age, ethnicity, family type)
Coefficient
Significance
No health insurance1
1.56
<.00001
4.8
3.1-7.4
Household income
<poverty level2
-.58
.001
0.6
0.5-0.8
<.01
.0001
1.8
2.4
1.2-2.8
1.5-3.7
Food insecure w/o hunger3
Food insecure w/ hunger3
Reference categories:
1
currently insured
2
100-199% poverty
3
food secure.
.60
.87
Odds Ratio
95% CI
Table II-xiii. Predictors of Delaying or Not Getting Care for Diabetes, among Patients Taking Oral
Hypoglycemic Agents, because Could Not Afford It
(Model adjusted for gender, age, ethnicity, household income as % of poverty, family type)
Coefficient
Significance
No health insurance1
1.99
<.0005
7.3
2.5-21.2
Food insecure with hunger2
1.62
<.0005
5.1
2.0-12.7
1
2
Referent is insured.
Referent is food secure.
Odds Ratio
95% CI
Table II-xiv. Predictors of Delaying or Not Getting Care for Arthritis, because Could Not Afford It
(Model adjusted for gender, age, family type)
Coefficient
No health insurance1
1.28
African-American2
-.78
Household income
< poverty level3
Food insecure
without hunger4
with hunger4
1
Reference is insured
Reference is white
3
Reference is 100-199% poverty
4
Reference is food secure.
2
Significance
<.00001
Odds Ratio
95% CI
3.6
2.3-5.5
.01
.45
.25-.83
-.38
<.05
.69
.48-.97
.64
1.03
<.005
<.00001
1.9
2.8
1.2-2.9
1.9-4.2
Table II-xv. Predictors of Delaying or Not Getting care for Heart Disease and Asthma, because Could Not Afford It
(Models adjusted for age, gender, ethnicity household income as % of poverty, and family type)
Coefficient
Heart disease
No health insurance1
10.3
95% CI
4.6-23.0
- .71
.01
.51
.27-.97
Food insecure
with hunger3
.87
<.05
2.4
1.2-4.8
1.17
<.00001
3.2
1.9-5.5
.64
.86
<.05
<.005
1.9
2.4
1.1-3.3
1.3-4.2
Referent is insured
Referent is white
3
Referent is food secure.
2
<.00001
Odds Ratio
African-American2
Asthma
No health insurance 1
Food insecure
without hunger3
with hunger 3
1
1.33
Significance