Top Clin Nutr
Vol. 33, No. 3, pp. 247–258
c 2018 Wolters Kluwer Health, Inc. All rights reserved.
Copyright
ORIGINAL RESEARCH
Literacy’s Role in Health
Disparities
A Mediator of Race and Income With
Anemia in African American and
White Adults
Emily S. Shupe, MS; Ryan T. Pohlig, PhD;
Marie Fanelli Kuczmarski, PhD;
Alan B. Zonderman, PhD; Michele K. Evans, PhD
Literacy impacts diet quality and may play a role in preventing anemia. This study investigated
whether literacy mediates the relationships between race or poverty status and diet quality and anemia. Diet quality was evaluated using mean adequacy ratios for 1895 white and African American
adults from Healthy Aging in Neighborhoods of Diversity across the Life Span study. Anemia was
diagnosed by World Health Organization standards. Path analysis explored the influence of race
and poverty on anemia. Anemia was diagnosed in 223 participants. The synergistic effects of
poverty, race, and diet quality influence anemia. Literacy mediated the effects of race and poverty
on mean adequacy ratios and anemia, highlighting the role of literacy in physical well-being.
Key words: African American, anemia, diet quality, health disparities, inflammation, iron
deficiency, literacy
A
CCORDING to the National Health and
Nutrition Examination Survey data, the
prevalence of anemia in the United States rose
Author Affiliations: Departments of Behavioral
Health and Nutrition (Ms Shupe and
Dr Kuczmarski) and Health Sciences (Dr Pohlig),
University of Delaware, Newark; and Laboratory of
Epidemiology & Population Sciences, National
Institute on Aging, National Institutes of Health,
Baltimore, Maryland (Drs Zonderman and Evans).
This work is supported by the Intramural Research Program, National Institute on Aging, National Institutes
of Health, grant Z01-AG000194.
The authors have disclosed that they have no significant relationships with, or financial interest in, any
commercial companies pertaining to this article.
Correspondence: Emily S. Shupe, MS, Department
of Behavioral Health and Nutrition, University of
Delaware, 206B McDowell Hall Newark, DE 19716-.
DOI: 10.1097/TIN-
from 4% to 7.1% from 2003 to 2012, with an
average of 5.6% of adults meeting the criteria for anemia.1 While 5.6% would indicate
that anemia is only of mild public health concern (World Health Organization citation),
anemia can be a serious health concern for
specific groups such as African Americans
(AAs), adults older than 60 years, nonpregnant women of reproductive age, and pregnant women. National Health and Nutrition
Examination Survey data indicate that anemia
is more prevalent in AAs than in whites (Ws)
(14.9% and 4%, respectively) and in females
than in males (7.6% and 3.5%, respectively),
with prevalence peaking at ages 40 to 49 years
of age and again at 80 to 85 years of age.
Among US adults, the prevalence of anemia
appears evenly distributed among nutritional
anemia (NA), anemia of inflammation (AI)
and unexplained anemia (UA).2 Nutritional
anemia is typically the result of deficiencies
247
Copyright © 2018 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
248
TOPIC IN CLINICAL NUTRITION/JULY–SEPTEMBER 2018
of iron, folic acid, and/or vitamin B12 .2
Anemia of inflammation is most commonly
associated with underlying conditions such
as atherosclerosis, diabetes mellitus, malignancy, and rheumatologic disorders.3–5 While
research on the demographic statistics of
individuals classified as either NA or AI is
limited, some research suggests that AAs tend
to have AI while Ws tend to have NA.2,6,7
Given the possible etiologies associated with
UA,8,9 UA was not included in this analysis.
To reduce the prevalence of anemia, several factors may contribute to its development
and should be explored. Race, poverty status,
and literacy are among the factors that contribute to health disparities.10,11 The higher
prevalence of anemia in AAs than in Ws could
be due to several factors. A 2014 report by the
US Census Bureau found more AAs than Ws
are uninsured (22% and 13.7%, respectively),
more AAs live in poverty than Ws (26.2% and
10.1%, respectively), and AAs have a lower
median income than Ws.11 African Americans
have higher rates of many of the leading
causes of death including, hypertension, diabetes, heart disease, stroke, and cancer.10
Those in the lowest income groups, compared with those with higher incomes, are
at a greater risk for developing anemia.12,13
Iron deficiency anemia is the most common
anemia among low-income groups.14 Individuals in the lowest income group (<100% of
the federal poverty level) are almost 5 times
as likely to have poor health as those in the
highest income group (≥400% federal poverty
level) and are almost twice as likely to have diabetes and heart disease,15 both of which may
be associated with developing anemia.5,16
Literacy and the discordance in estimates
of literacy levels of individuals by physicians
and other health professionals have been suggested as important contributors to health
disparities.17,18 Low literacy is significantly
associated with poor overall health status,
more hospitalization, and greater rates of
mortality.19–21 African Americans and other
minority races are more likely to have
marginal or low literacy skills.22 Individuals
with low incomes also tend to indicate lower
health literacy compared with those with
higher incomes.22 Understanding the role of
literacy among racially and economically diverse populations may be critical to reducing
the disparities of health conditions such as
anemia.
Literacy appears to contribute to the dietary differences observed between AAs and
Ws, with AAs having lower diet quality, resulting in a greater need to improve diet quality to
optimize health.23,24 In addition, low-income
adults generally have poorer diet quality than
higher-income adults.25,26 Dietary intake may
contribute to the development of anemia, either via energy insufficiency and inadequate
micronutrient intake (NA)27 or energy imbalance of macronutrients leading to inflammatory conditions that contribute to the development of AI.28,29
The exploration of literacy and diet quality
on the presence of anemia in a racially and
socioeconomically diverse population might
be beneficial for the prevention and intervention efforts. The Health Aging in Neighborhoods of Diversity across the Life Span (HANDLS) study was designed to investigate the effects of race and socioeconomic status (SES)
on health disparities.30 Although the primary
health outcomes were cardiovascular disease
and cognitive function, comprehensive medical, dietary, and demographic data were collected. Using data from this study, the aims of
this project were to investigate (1) the effects
of race and poverty status on the presence
of NA and AI and (2) the roles that literacy
and diet quality play in the complex relationship between race and poverty status and the
presence of NA or AI.
METHODS
HANDLS study
The HANDLS study is a prospective
population-based cohort study. The baseline
cohort -) comprised 3720 AA and
W adults 30 to 64 years of age. Participants
were drawn from 13 predetermined neighborhoods (defined as groups of contiguous
census tracts) in the city of Baltimore,
Copyright © 2018 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Literacy’s Role in Health Disparities
Maryland. Four factors were selected for
sampling: age (7, 5-year age bands from 30
to 64 years), self-reported race (AA and W),
sex, and SES (operationalized by household
incomes below or above 125% of the 2004
federal poverty guidelines). Those participants with household income above 125%
of the poverty line were identified as above
poverty and those below as below poverty
(BP). The HANDLS baseline population had
a mean age of 47.7 years, 59% were AA, 45%
were male, and 41% of participants were
BP.30 All participants spoke English. Individuals were excluded if they were pregnant
or within 6 months of active treatment for
cancer. Further detailed information on the
study design, eligibility, and recruitment of
participants can be found elsewhere.30
The HANDLS baseline data were collected
in 2 phases conducted 4 to 10 days apart.31
Phase 1 was a household survey conducted in
the participant’s home by a recruitment and
sampling contractor. Participants completed
questionnaires detailing their background,
demographic information, education experience, occupational status, household income,
physical activity, and health status information. A trained interviewer administered the
first 24-hour dietary recall. The second phase
took place in mobile research vehicles, where
a medical history survey, a physical examination, fasting blood draw, cognitive testing,
and body composition measurements, and
physical performance measures were administered by a physician or member of medical
staff. A trained interviewer administered a
second 24-hour dietary recall.
Study population
Of the 3720 eligible participants enrolled
at baseline, 2177 participants completed both
24-hour dietary recalls. Of the 2177 with complete dietary data, 89 participants were excluded for having chronic kidney disease or
sickle cell disease and 111 were excluded
for missing blood markers of anemia (serum
ferritin, transferrin saturation percentage [FeSat], serum folate, and serum vitamin B12 ).
Anemia criteria were applied to the remaining
249
1977 eligible participants. Participants with
UA were excluded (n = 82). No other exclusions were made. Institutional review boards
at National Institute of Environmental Health
Sciences and at the University of Delaware approved the study protocol. All HANDLS participants provided written informed consents
and were compensated monetarily. A flow
diagram of the household sampling to eligible participants for this study is presented in
Figure 1.
Dietary methods
The United States Department of Agriculture (USDA) Automated Multiple Pass Method
was used to conduct both 24-hour dietary
recalls. Recalls were conducted on all days
of the week to account for variations in
diet between weekdays and weekends. The
Automated Multiple Pass Method involves
5 steps designed to provide cues and prompt
thorough recall for all foods and drinks consumed throughout the previous day.32 These
steps include (1) quick list of all foods consumed the previous day; (2) a forgotten foods
list that includes probes for commonly forgotten foods; (3) probes to determine the time
a food was consumed and at which meal; (4)
detailed questions including amounts of foods
consumed, additions to foods, and where food
was obtained; and (5) a final review to probe
any foods not previously remembered. An illustrated food model booklet as well as other
visual measurement aids was used to increase
the accuracy of estimating food and drink
quantities. A trained coder using the USDA
Survey Net data-processing system to match
the foods with codes in the Food and Nutrient Database for Dietary Studies version 3.0
coded each recall.33 The nutritional supplement questionnaire was not administered during baseline data collection to minimize subject burden since the participants were on the
mobile research vehicles for at least 6 hours.
Clinical measures
Participants were weighed (kg) without
shoes and coats using a calibrated Health
O Meter digital scale (Pelstar, LLC, Alsip,
Copyright © 2018 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
250
TOPIC IN CLINICAL NUTRITION/JULY–SEPTEMBER 2018
Figure 1. A flow diagram of the household sampling to eligible participants for this Healthy Aging in
Neighborhoods of Diversity across the Life Span study. CKD indicates chronic kidney disease; HANDLS,
Healthy Aging in Neighborhoods of Diversity across the Life Span; HIV, human immunodeficiency virus.
Illinois). Height (cm) was obtained with
the participant’s heels and back against a
height meter supplied by Novel Products, Inc
(Rockton, Illinois), and body mass index
(BMI) was calculated from measured height
and weight. Body mass index was calculated using the equation: BMI = weight
(kg)/[height (m)]2 .31 For this project, participants were categorized as having a BMI of
less than 25 (normal) or a BMI of 25 or more
(overweight and obese).
Fasting venous blood specimens were collected from participants in the morning when
they arrived at the mobile research vehicles
visit and analyzed by Quest Diagnostics, Inc
(Chantilly, Virginia). Fasting blood results included 2 indicators for iron (Fe) status: FeSat
(transferrin saturation) (serum iron/total iron
binding capacity) and serum ferritin (ng/mL),
plus measures of serum folate (ng/mL), serum
vitamin B12 (pg/mL), and C-reactive protein
(CRP) (mg/L). Serum Fe and total iron binding capacity were assessed using the standard
clinical laboratory spectrophotometric assay.
Serum ferritin was measured using a standard
chemiluminescence assay. Serum folate and
vitamin B12 were measured using enzyme immunoassay. High-sensitivity CRP levels were
assessed by the nephelometric method utilizing latex particles coated with CRP monoclonal antibodies. All assays were considered
reliable but with any method there are limitations that are described on the Quest diagnostics Web site.34
Assessment of anemia
Participants with anemia were identified by
the World Health Organization standard that
Copyright © 2018 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Literacy’s Role in Health Disparities
defines anemia as a Hb level less than 13 g/dL
in men and less than 12 g/dL in women.35
After identifying individuals with anemia, the
type of anemia was determined.36,37 Anemic
individuals were classified with NA if they had
low serum ferritin (≤30 ng/mL) or normal
serum ferritin (31-99 ng/mL) and as low FeSat
(<16%) if they had folate less than 4 ng/mL or
if they had vitamin B12 less than 200 pg/mL.37
Anemia of inflammation was defined as having elevated serum ferritin (≥100 ng/mL) or
normal serum ferritin (31-99 ng/mL) with elevated FeSat (>45%).
Measures
Variables included in the analyses were
race, poverty status, literacy, and diet quality.
Race was self-reported as AA or W. Poverty
status was defined as self-reported household
income dichotomized into less or greater than
125% of the 2004 federal poverty guidelines.
Literacy was assessed using the reading subtest of the Wide Range Achievement Test
(WRAT)—3rd Edition.38 The WRAT measures
the ability to recognize and name letters and
words. The total WRAT Reading score equaled
the sum of total correctly pronounced letters
and total correctly pronounced words and
served as the literacy measurement. A WRAT
score of 37 to 40 represents a sixth- to eighthgrade reading level, and a score of 41 to 46
represents a high school reading level.
Nutrient adequacy ratio (NAR) and mean
adequacy ratio (MAR) scores were used to assess nutrient-based diet quality.39,40 The NAR
score was determined by taking each participant’s daily intake of a nutrient divided by the
recommended dietary allowance for that nutrient. Nutrient adequacy ratio scores were
determined for 17 micronutrients: vitamins
A, C, D, E, B6 , B12 , folate, iron, thiamin, riboflavin, niacin, copper, zinc, calcium, magnesium, phosphorus, and selenium. The recommended dietary allowance was adjusted
for the age and sex of participants and vitamin C was adjusted for smokers.41 The
NAR score was then converted into a percent with values exceeding 100 truncated to
100. Mean adequacy ratio scores were calcu-
251
lated
by averaging the NAR scores: MAR =
( NAR scores)/17.40 Nutrient adequacy ratio and MAR were calculated separately for
each daily intake and then averaged together.
Mean adequacy ratio scores, based on food
intakes only, were used as the nutrient-based
diet quality variable.
Covariates
Demographic characteristics included age
(years) and sex (male or female). Cigarette
smoking was coded as current smoker or nonsmoker. Handgrip strength, which is used as
an indicator of total-body muscle strength and
physical performance, was assessed using the
Jamar Hydraulic Hand Dynamometer.42 The
hand dynamometer registers the maximum
kilograms of force per trial. Two trials were
conducted with both the right and left hands.
The mean of the 2 trials by the dominant hand
was used in the analysis.
Statistical analyses
Demographic and descriptive statistics
means and standard errors for continuous variables and frequencies and percentages for categorical variables are reported in the Table.
One-way analyses of variance for continuous
variables and χ 2 tests for categorical variables
were used to test for differences among participants with NA, participants with AI, and participants not diagnosed with anemia. Two statistical programs were used for analysis: SPSS
software, release 23 and Mplus software, version 7.31. For the path analysis, standardized
estimates are reported. Alpha level was set
to .05.
Path analysis is a special case of Structural
Equation Modeling in which all variables in
the model are directly observed. It is a multivariate statistical technique that allows researchers to test complicated models, while
allowing variables to be both outcomes and
predictors at the same time. Conceptually, it
can be considered a set of simultaneously estimated multiple regression equations.
Approaching this analysis through multiple
regressions rather than path analysis would
be burdensome as it would require a separate
Copyright © 2018 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
252
TOPIC IN CLINICAL NUTRITION/JULY–SEPTEMBER 2018
Table. Characteristics of Healthy Aging in Neighborhoods of Diversity Across the Life Span
Study Participants Categorized by Anemia Diagnosis by World Health Organization Criteriaa
Nutritional
Anemia
(N = 160)
Sex; % female
Race; % AA
Income, %; <125% poverty
Current smoker, %
BMI ≥25 kg/m2 , %
Age, X̄ ± SE, y
WRAT score, X̄ ± SE
MAR, X̄ ± SE
Handgrip strength, X̄ ± SE, kg
CRP mg/L, X̄ ± SE
Energy, X̄ ± SE, kcal
Hb, X̄ ± SE, g/dL
Serum iron, X̄ ± SE, μg/dL
Serum ferritin, X̄ ± SE, ng/mL
FeSat, X̄ ± SE, %
Folate, X̄ ± SE, ng/mL
Vitamin B12 , X̄ ± SE, pg/mL
a
85.6
80.6a
51.9a
38.5a
75.6
43.8 ± 0.7a
41.3 ± 0.6a,b
67.3 ± 1.2a
30.9 ± 0.7a
8.4 ± 1.0a
1847 ± 80
11.0 ± 0.1a
45.8 ± 3.0a
21.0 ± 1.8a
11.8 ± 0.6a
12.8 ± 0.4a
480.3 ± 17.1a
Anemia of
Inflammation
(N = 63)
b
42.9
90.5a
47.6a,b
55.6b
67.7
52.8 ± 1.1b
39.4 ± 1.1a
72.3 ± 1.5a,b
35.5 ± 1.5a,b
15.0 ± 4.3b
2061 ± 134
11.6 ± 0.1b
73.5 ± 5.5b
226.4 ± 19.1b
23.1 ± 1.7b
13.7 ± 0.9a,b
585.5 ± 38.9b
No Anemia
(N = 1672)
b
53.8
52.0b
40.8b
49.1b
70.8
47.8 ± 0.2c
42.5 ± 0.2b
71.8 ± 0.4b
35.6 ± 0.5b
4.3 ± 0.2c
2032 ± 24
14.1 ± 0.03c
88.9 ± 0.9c
129.6 ± 4.0c
26.0 ± 0.3b
14.5 ± 0.2b
512.7 ± 5.7a
P