Family Economics and Nutrition Review [Volume 13, Number 2]

              CENTER FOR NUTRITION POLl('( AND PROMOTION
Development of a Food Checklist for Fat, Saturated
Middle School Students
Johanna T Dwyer. Anne 0. Garceau, Deanna M. Hoelscher, Kevin W Smith.
Theresa A. Nicklas, Leslie A. Ly tle. Michelle M. Zive, and Ann L. Clesi
12 Using Credit to Cover Living Expenses: A Profile of a Potentially Risky Behavior
Gabriela Castellani and Sharon A. DeVaney
21 Relationships of Substance Abuse to the Nutritional Status of Pregnant
African-American Women
Hazel A.B. Hiza, Allan A. Johnson, Enid M. Knight, Claudette S. Welch,
and Cecile H. Edwards
Influences on Fruit and Vegetable Procurement and Consumption Among
Urban African-American Public Housing Residents, and Potential Strategies
for Intervention
Sharada Shankar and Ann Klassen
Research Briefs
47 Caffeine and Theobromine Intakes of Children:
Jaspreet K. C. Ahuja and Betty P Per/off
52 Insight 16: Body Mass Index and Health
Hazel A. Hiza, Charlotte Pratt, Anne L. Mardis. and Rajen Anand
55 Insight 17: A Look at the Diet of Pregnant Women
Anne L. Mardis and Rajen Anand
58 Insight 18: Food Insufficiency and the Nutrit ional Status of the Elderly Population
Nadine Sahyoun and P Peter Basiotis
Regular Items
USDA Activities •
Consumer Prices
UNITED STATES DEPARTMENT OF AGRICULTURE
Volume 13, Number 2
2001
.... , .....
,· ..
' "
Ann M. Veneman, Secretary
U.S. Department of Agriculture
Eric M. Bost, Under Secretary
Food, Nutrition, and Consumer Services
Steven N. Christensen, Acting Deputy Director
Center for Nutrition Policy and Promotion
P. Peter Basiotis, Director
Nutrition Policy and Analysis Staff
The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and
activities on the basis of race, color, national origin, gender, religion, age, disability, political
beliefs, sexual orientation, or marital or family status. (Not all prohibited bases apply to all
programs.) Persons with disabilities who require alternative means for communication of
program information (Braille, large print, audiotape, etc.) should contact USDA's TARGET
Center at (202) 720-2600 (voice and TDD).
To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room
326-W, Whitten Building, 14th and Independence Avenue, SW, Washington, DC 20250-
9410 or call (202) 720-5964 (voice and TDD). USDA is an equal opportunity provider and
employer.
_'
Ed tor
Ju a M. Dinkins
As istant Editor
De vid M. Herring
Fe ttures Editor
M< rk Lino
M· naging Editor
Jo e W. Fleming
Pt> H Review Coordinator
H< zel Hiza
Fo 1ily Economics and Nutrition Review is
w1 ten and published each quarter by the
Ct 1ter lor Nutrition Policy and Promotion,
U. Deportment of Agriculture, Washington,
DC
Th ' Secretory of Agriculture hos determined
th t publication of this periodical is necessary
ir he transaction of the public business
rf !Uired by law of the Department.
n I publication is not copyrighted. Contents
m< y be reprinted without permission, but credit to
Fe nily Economics and Nutrition Review would
lx appreciated. Use of commercial or trade
nt nes does not imply approval or constitute
e1 Jarsement by USDA. Family Economics and
N trition Review is indexed in the following
d taboses: AGRICOLA, Ageline, Economic
lt •rature Index, ERIC, Family Studies, PAIS,
0 1 d Sociological Abstracts.
F, 7lily Economics ond Nutrition Review is lor
S< .e by the Superintendent of Documents.
S bscription price is $19.00 per year ($23.75
k foreign addresses). Send subscription order
a d change of address to Superintendent of
C >euments, P.O. Box 371954, Pittsburgh, PA
1 250-7954. (See subscription form on p. 79.)
C iginol manuscripts are accepted lor
P blication. (See "guidelines lor authors" on
b ck inside cover.) Suggestions or comments
c. ncerning this publication should be addressed
to Julio M. Dinkins, Editor, Family Economics
or :1 Nutrition Review, Center lor Nutrition Policy
or ~ Promotion, USDA, 31 01 Park Center Drive,
Rcom 1034, Alexandria, VA 22302-1594 .
Tha Family Economics and Nutrition Review
is now available at http://www.cnpp.usdo.gov.
(S ~ e p. 78 .)
o; 6! ;port
3 Development of a Food Checklist for Fat, Saturt'~Sba~ ~d Sodium for
Middle School Students 'Oro ji"OqiJa-
Johanna T. Dwyer, Anne 0. Garceau, Deanna M. Hoelscher, Kevin W Smith,
Theresa A. Nicklas, Leslie A. Lytle, Michelle M. Zive, and Ann L. Clesi
12 Using Credit to Cover Living Exp~nses : A Profile of a Potentially Risky Behavic
Gabriela Castellani and Sharon A. DeVaney
21 Relationships of Substance Abuse to the Nutritional Status of Pregnant
African-American Women
Hazel A.B. Hiza, Allan A. Johnson, Enid M. Knight, Claudette S. Welch,
and Cecile H. Edward~
34 Influences on Fruit and Vegetable Procurement and Consumption Among
Urban African-American Public Housing Residents, and Potential Strategies
for Intervention
Sharada Shankar and Ann Klassen
Research Briefs
4 7 Caffeine and Theobromine Intakes of Children : Results From CSFII 1994-96, 1998
Jaspreet K. C. Ahuja and Betty P Per/off
52 Insight 16: Body Mass Index and Health
Hazel A. Hiza, Charlotte Pratt, Anne L. Mardis, and Rajen Anand
55 Insight 17: A Look at the Diet of Pregnant Women
Anne L. Mardis and Rajen Anand
58 Insight 18: Food Insufficiency and the Nutrit ional Status of the Elderly Population
Nadine Sahyoun and P Peter Basiotis
Regular Items
6 1 Resea rch and Evaluation Activities in USDA
64 Federal Studies
7 2 Journal Abstracts
7 4 Official USDA Food Plans: Cost of Food at Home at Four Levels,
U.S. Average, October 2001
7 5 Consumer Prices
7 6 U.S. Poverty Thresholds and Related Statistics
7 7 Reviewers of 2001 Articles
Volume 13, Number 2
2001
CALL FOR PAPERS
SPECIAL IssuE oN ELDERLY NuTRITION
The
Family Economics and Nutrition Review
is calling for manuscripts
for a special issue on the Elderly.
This issue will highlight nutrition for the elderly but will also emphasize physiological, socio-economical,
familial, psychological, environmental, educational, and other factors that influence the nutritional status of
older Americans. Authors are invited to submit papers on basic or applied research, innovative research
approaches, methods, education programs, and economic issues that focus on improving nutritional health
or policy issues for older Americans. Research articles and research briefs will be considered.
Original manuscripts must be postmarked by Monday, April 1, 2002.
When submitting articles, see guidelines for authors on the CNPP Web site:
www. cnpp. usda.gov
Johanna T. Dwyer, DSc, RD
Tufts University School of
Medicine and Nutrition
Frances Stern Nutri tion Center
Anne 0. Garceau, MS, RD
Frances Stern Nutrition Center
New England Medical Center
Deanna M. Hoelscher, PhD, RD
Center for Health Promotion Research
and Development
University of Texas-Houston
Kevin W. Smith, MA
New England Research Institute
Theresa A. Nicklas, DrPH, LON
Tulane Center for Cardiovascular
Health
Tula ne University School of
Public Health and Tropical Medicine
Leslie A. Lytle, PhD, RD
School of Public Health
University of Minnesota
Michelle M. Zive, MS, RD
Department of Pediatrics
University of California
Ann L. Clesi, MEd
Tulane University School of
Public Health and Tropical Medicine
2001 Vol. 13 No.2
Research Articles
Development of a Food Checklist
for Fat, Saturated Fat, and
Sodium for Middle School
Students
We developed a brief, inexpensive, culturally sensitive 24-hour food checklist
to identify middle school students enrolled in the Child and Adolescent Trial
for Cardiovascular Health (CATCH), whose food choices over the previous day
were high in total fat, saturated fat, or sodium. Food checklists were coded
from 224 24-hour recalls previously collected from CATCH students in the
fifth grade to simulate responses to it. Administration procedures for the food
checklist were then pretested on 71 schoolchildren in grades 6 through 8.
Regression results indicated that consumption of 1 0 items on the checkl ist had
a positive effect on fat intake; 13, on saturated fat; and 11, on sodium intake.
Some foods were removed from the checklist because of their small effect size
or infrequency of reported consumption; others were combined or subdivided
to form new food groups, or were reworded to improve comprehension. The
final food checklist consisted of 40 foods or food groups. The median same­day
test-retest reliability Kappa was 0 .85; item validity, as measured by the
median Kappa statistic, was 0.54 . The food checklist procedures described
may be helpful for developing similar food checklists. Nutrition educators
and teachers may find that the food checklist is a useful educational tool for
informing students about their fat intakes.
Although 24-hour recalls and food
records are the most accurate of
the dietary assessment methods
available, they require a great deal of
instruction and are too expensive and
burdensome to use in large-scale
community studies (28). Thus brief,
inexpensive, valid, culturally appropri­ate
dietary assessment instruments that
can be used to categorize children's
relative intakes of nutrients are needed.
Food checklists are useful in large­scale
studies for detecting changes in
food choices and for quantifying and
ranking individuals' intakes of specific
nutrients. When used to assess the prior
day's food consumption, food check­lists
can be calibrated by comparing
results with 24-hour recalls.
The Child and Adolescent Trial for
Cardiovascular Health (CATCH) was
a large-scale, school- and family-based,
multicenter intervention trial aimed at
decreasing cardiovascular risk factors
and making organizational-level
changes. The cohort consisted of
elementary schoolchildren and their
schools. Details of the CATCH study
are described elsewhere (21). Particular
attention was directed at educating
children on positive eating behaviors
to improve and lower intakes of
sodium, total fat, and saturated fat (19) .
A food checklist was designed as a
short, inexpensive diet assessment
tool to detect differences between the
target nutrients in the diets of the
3
CATCH cohort as they were followed
longitudinally. Use of a checklist
appeared promising, but existing
checklists were either inappropriate
for children or the targeted nutrients
differed from those of interest in
CATCH. For example, Krista) and
colleagues (I 4) used a 19-item checklist
of foods high in fat and fiber-although
neither saturated fat nor sodium was
included-to study women's intakes.
For 16 foods, Kappa values exceeded
0.6 when food items reported on the
checklist and 24-hour recalls were
compared.
In students followed from the sixth
through the twelfth grades in the Class
of '89 Study, an 18-item scale or paired
food choice was used (I 3). This scale
detected differences in high-fat food
choices between students residing in
intervention and control communities,
and scores suggestive of a preference
for high-fat foods correlated well
with lack of exercise and smoking
(12,13,17,20). Middle school students
in the CATCH intervention group
differed from controls on their usual
choices between food pairs on the
Health Behavior Questionnaire (I 6) .
However, items on the Class of '89
and the CATCH Health Behavior
Questionnaire food-choice scale
asked subjects to indicate which of
two food pairs they usually choose
rather than asking them to report their
food consumption.
The purpose of this study was to
develop a brieffood checklist to
report intakes of foods that were major
contributors to middle school children's
intake of fat, saturated fat, or sodium
over the previous day. This checklist
needed to be inexpensive, culturally
sensitive, and suitable for administra­tion
in group settings to a multiethnic
group of middle and junior high school
students.
4
Methods
Sample
The data we used consisted of 224
24-hour recalls, 56 at each of the
four CATCH sites: San Diego, CA;
New Orleans, LA; Minneapolis, MN;
and Austin, TX. These recalls were
selected randomly from all 1,182
recalls collected in the CATCH study
after stratifying by site when the cohort
children were in the fifth grade, in
the spring of 1994. The multiethnic
sample reflected the composition of
the CATCH population: 44 percent
females and 56 percent males; 68
percent White, 13 percent Black, 15
percent Hispanic, and 4 percent Native
American, Asian American, and others.
The sample size was selected to permit
us to detect reliably food-item effect
sizes of 0.35 or greater (8), sizes we
considered large enough to be of
dietary importance. Effect size is the
difference in mean nutrient intake levels
between those who consumed a food
versus those who did not, divided by
the standard deviation of the measured
nutrient.
Preliminary List of Foods
Included on the Food Checklist
Developing the food checklist involved
(1) compiling the preliminary list of
foods to be included on the food
checklist, (2) coding the food checklist
by using previously obtained 24-hour
recall information to simulate student
response, (3) calibrating the food
checklist to 24-hour recalls to produce
a final version of it for administration,
and (4) formalizing administration
procedures after pretesting the food
checklist administration with students.
The food checklist was a modification
of a questionnaire used in the Youth
Risk Behavior Survey (I 1), with some
items from the Food Behavior Checklist
(14). It included food choices, reported
by third graders in the CATCH pilot
study, that were high in fat, saturated
fat, and sodium and similar foods
identified in other studies (3-5, 10,
26,30). Other questionnaires that
focused on fat, saturated fat, and
sodium were also reviewed even if
they were not designed for children or
adolescents (1 ,2,6,9,11,14,24,25,29) .
Foods that contributed substantially to
intakes of target nutrients, because they
were consumed frequently, were also
included (5,26,30). In addition, special
attention was paid to inclusion of
relevant ethnic foods.
To cluster foods into groups that were
similar in their nutrient composition,
we examined tables that reported food
composition for total fat, saturated fat,
and sodium based on nutrients per 1 00
grams of each food item. Items consid­ered
by themselves and items added to
foods, such as butter and salad dressing,
were considered for inclusion. When­ever
possible, foods and groups of
foods were categorized similarly to
those employed on existing instruments,
such as on the food frequency question­naire
of the Third National Health
and Nutrition Examination Survey
(NHANES III) (22) and the Block
Brief Fat Screener (2).
Analysis of Existing 24-Hour
Recall Records
Foods to be included in the checklist
were evaluated by using a criterion­oriented
approach similar to that
described by Posner and colleagues
(23). For this analysis, a random
subsample of 224 of the records
collected in 1994 from fifth grade
CATCH students using 24-hour recalls
was used. These recalls were collected
by trained and certified CATCH
interviewers who used standard
techniques and a nonquantified food
record as a memo prompt during their
Family Economics and Nutrition Review
recalls (18). The Minnesota Nutrition
Data System (NDS) software 1 was used
to compute nutrient intakes. The food
records were listed on the checklist,
examined, and scored "yes" if the
food was eaten and "no" if it was not.
Foods on the recall that were not on the
checklist were not scored. The criterion
was the extent of agreement between
the score on the food checklist and
the intake of a specific nutrient (fat,
saturated fat, or sodium) on the 24-hour
recall of the previous day as assessed
from stepwise linear regression.
For this study, copies of the previously
collected 24-hour recalls were obtained
for each of the students. Five nutrition­ists
(one at each of the four CATCH
field sites and one at the Data Coordi­nating
Center) were trained to use
standardized written instructions to
code the food checklist from 24-hour
recall printouts and followed these
procedures. The site nutritionists then
completed a food checklist for each of
56 recalls collected at each site, and the
Data Coordinating Center nutritionist
filled out a total of 56 food checklists
( 14 collected at each site). The nutri­tionists
reviewed each food on the
recall and then marked the correspond­ing
item on the food checklist, thereby
providing a simulated checklist data
set for study purposes.
To evaluate inter-coder reliability,
another Data Coordinating Center
nutritionist coded all 224 recalls onto
checklists (table I). Agreement between
the coding of the quality assurance
coder and that of the five nutritionists
was tested with a generalized Kappa
statistic. Kappa values of 0.6 or greater
are generally regarded to indicate
"substantial" levels of inter-rater
agreement (1 5). Kappa values ranged
1The software was developed by the Nutrition
Coordinating Center (NCC), University of
Minnesota, Minneapolis, MN (Food Database
version 4a; Nutrient Database version 19).
2001 Voi.I3No.2
from 0.4 to 1.0, with 38 items exceed­ing
0.6 Kappa values. Inter-coder
reliability was excellent, with Kappas
usually exceeding 0.9 for individual
food groups, indicating nearly perfect
agreement in coding between the
quality assurance nutritionist and the
five nutritionists (site and Data Coordi­nating
Center). It was recognized that
nutrient correlations between children's
24-hour recalls and checklists coded
by nutritionists were higher than would
be found if children had completed the
food checklist. In actual use, children
may forget foods they have eaten or
misinterpret checklist items.
Pretest of the Administration
of the Food Checklist
The food checklist was administered,
by using a standardized protocol, to a
total of 71 nonrandomly selected sixth­(
n=l), seventh- (n=60), and eighth-
(n= 1 0) grade students representative of
the ethnic groups in the CATCH, with
nearly equal numbers of males and
females in seven groups at three of the
sites (California, Louisiana, and Texas).
Students were instructed to circle "yes"
next to any food group or food from
which they had consumed at least one
bite or one sip on the previous day.
They were instructed to categorize
unlisted foods such as sandwiches by
their separate components (e.g., bread,
ham, cheese, butter), and to circle all
the items that applied.
Following the administration of the
food checklist, the students were asked
a predesignated series of questions by
a food checklist administrator to assess
their understanding of instructions and
the clarity of items on the checklist.
Feedback from the students' observa­tions,
suggestions of the checklist
administrators, and the recommenda­tions
of the CATCH Dietary Assess­ment
Working Group were used to
revise the protocol used to administer
the food checklist.
Statistical Methods
Using the 224 recalls and their corre­sponding
food checklists, we conducted
a stepwise linear regression analysis
to evaluate the ability of items on the
food checklist to explain the variance
in nutrient levels obtained with the 24-
hour recalls. Individual food items were
assigned a " 1" if the item was checked
on the food checklist and a "0" if it was
not. Regression analyses were used to
determine the relative contributions of
individual foods or groups of foods on
the food checklist (independent or
predictor variables) to nutrient intakes
from 24-hour recalls for each of the
dependent variables (e.g., fat, saturated
fat, and sodium). The regression
coefficients were then converted to
effect sizes (regression coefficient
divided by the standard deviation for
each nutrient) (8).
Foods or food groups with small
effect sizes (i .e., less than 0.20) for
each target nutrient and those reported
infrequently (by less than 2 percent
of the students) were reviewed by the
CATCH Dietary Assessment Working
Group. Some items with small effect
sizes, such as bread and cookies, were
retained based on their high frequency
of consumption or status as major
contributors of target nutrients as
indicated in other studies. Otherwise,
such items were eliminated from the
checklist. The food items and food
categories on the food checklist were
edited for readability. Further testing
and validation studies as well as
additional details on scoring are
described in greater detail elsewhere
(27).
Results
The most commonly eaten foods were
bread, cookies, cold cereal, and potato
chips-all eaten by more than 44
percent of the children on the recall
day (table I).
5
Table 1. Final food checklist items with frequency of consumption, inter-coder reliability, and effect sizes on 224
middle school children
Food Effect sizes3
checklist Percent
number Percent Percent Kcal from Sodium
(final Food of children Kappa Kcal saturated (mg per
list) category eating item1 values2 from fat fat 1,000 kcal)
22 Bread 78 0 .85
29 Cookies 54 0.95
23 Cold cereal 49 0.99
26 Potato chips 44 0.99 0.28
14 Cheese 36 0.92 0.37 0.32
2 Hamburgers 34 0.92 0.49 0.43
39 Ketchup 33 0.94
3 Fried chicken 30 0.97 0.4 1
17 2% fat milk 30 0.95 0.42 0.56
21 Biscuits 30 0.97
31 Ice cream 27 0.94 0.62
19 French fries 26 0.99 0.24
32 Chocolate candy 26 0.96 0 .28
16 Whole milk 25 0.95 0.47 0 .69
7 Cold cuts 24 0.93 0.54
33 Margarine 23 0.90
35 Mayonnaise 21 0.93 0.25 •
12 Piua 19 0 .96 0.37 0.33
28 Peanut butter 16 0.98 0.38
8 Bacon 12 0.83 0.95 0 .55 0.58
27 Pickles 11 1.00 0.59
11 Spaghetti with meat sauce 11 0.95
34 Butter 10 0.88 0 .58
24 Pancakes 9 0.83
13 Cheese dishes4 8 0.87 0.45
10 Soup 8 1.00 0 .86
18 Beans5 8 0.66
1 Beef 7 0.70
6 Hot dogs 7 0.97 0 .66 0 .62 0.56
37 Gravy 7 0.97 0.56 0.34
4 Turkey 6 0.96 0.41
38 Whipped cream 6 0.96 0 .56
15 Eggs 5 1.00
20 Spanish rice 5 0.82 0.45
40 Salt 5 0.65 0 .58
5 Meat salad < 1 0.66
36 Salad dressings6
9 Pork 7 0 .97
25 Pretzels 4 1.00
30 Donuts6
• Effect size < 0.20.
-No scores are available because the item was originally port of another food group.
1 Site and Data Coordinating Center coding.
2 Quality assurance coding versus site and Data Coordinating Center coding.
3 Effect sizes were calculated by dividing the regression coefficient by the standard deviation for each nutrient.
4 Macaroni and cheese, cheese nachos, cheese enchiladas, quesadillas.
5 Red, white, baked, refried.
6 Salad dressings and donuts are included only to illustrate a! I items on the final checklist; these items were originally included in other groups.
Nate: Foods or food groups with Kappa values <0.60 are not shown in the table.
6 Family Economics and Nutrition Review
The effect sizes show that consumption
of 10 of the 40 food items/groups had
a positive influence on fat intake
expressed as percentage of calories.
Thirteen food items/groups had a
similar effect on percentage of calories
from saturated fat. And 11 food items/
groups had a similar effect on sodium
intake per I ,000 calories. Larger effect
sizes indicate a greater contribution to
target nutrient intakes on the 24-hour
recall. Effect sizes ranged from less
than 0.2 to 0.95 (bacon) for percentage
of kilocalories from fat; less than 0.22
to 0.69 (whole milk) for saturated fat,
and Jess than 0.2 to 0.86 (soup) for
sodium.
Twenty-one of the original 45 food
items/groups had minor-2 effects on
nutrient profiles for total fat, saturated
fat, or sodium levels. These were
examined further, and six items
(biscuits, bread, cold cereals, cookies,
margarine, and ketchup) with minor
effects on nutrient profiles were
retained on the checklist because they
were consumed by a substantial number
(23 to 78 percent) of the students. Six
other items (beef, pork, spaghetti with
meat sauce, eggs, and pretzels) were
retained because they made substantial
contributions to intakes of one or more
of the CATCH target nutrients reported
in other studies (3-5,10,26,30).
Meat salads (e.g., tuna, chicken, or
shrimp salad) and pancakes were
infrequently consumed and had minor
effect sizes but were retained because
the older middle and junior high school
students, the target population, would
likely consume these foods. Canned
beans (pork and beans and pinto beans)
were infrequently consumed but
retained because of their popularity
among Hispanic-American children.
2Standard deviation less than 0.2.
2001 Vol. 13 No. 2
Six items were deleted (canned
vegetables, mashed potatoes, granola,
trail mix, dips, and french toast). Two
food items/groups were recategorized
into groups that more adequately
reflected nutrient content. Cookies
were divided into two groups (cookies
and donuts) to narrow the range offat
content per 1 00 grams in each group.
In addition, barbecue sauce was
combined with ketchup with the
rationale that this regrouping might
reveal larger effect sizes in future
testing with older children.
On the food checklist pretest, instruc­tions
took about 5 minutes and the food
checklist took 10 minutes for students
to complete. Specific references to
Jowfat and low-sodium foods were not
included in the instructions to students,
because structured feedback with
students revealed that they were unable
to distinguish between lowfat, fat-free,
and regular food items. However, these
issues were discussed in directions to
the administrators. Students who ate a
lowfat or low-sodium version of a food
on the food checklist and asked the
administrators how to complete the
checklist were instructed to circle
"yes" next to the checklist item. A
list of commonly asked questions and
standard answers for administrators
was developed based on questions
encountered in the pretest
administration.
This developmental study was done
in preparation for a validation study,
which compared seventh grade
students' 24-hour recalls with check­lists
they completed the same day. The
purpose of this phase of the develop­ment
was to identify the appropriate
food items for the checklist. The
psychometric properties of the instru­ment
were tested after this process was
completed. These and other aspects of
the scoring and validation study are
reported in detail elsewhere (27) .
The most commonly eaten
foods were bread, cookies,
cold cereal, and potato
chips--all eaten by more than
44 percent of the children on
the recall day.
7
... the 40-item prototype food
checklist developed to serve as
a surrogate to the 24-hour
recall was feasible.
8
Briefly, the median same-day test­retest
reliability Kappa was 0.85, and
item validity-as measured by the
median Kappa statistic comparing
student choices with those of staff
nutritionists-was 0.54 (27). The final
food checklist items shown in table I
consisted of 40 items ( 4 single foods,
25 food categories, 2 beverages, 3
single condiments, and 6 condiment
groups).
Discussion
The major fmding of this study is that
the 40-item prototype food checklist
developed to serve as a surrogate to
the 24-hour recall was feasible. Nearly
half of the items on the original check­list
had no appreciable effects on
regressions for total fat, saturated fat,
and sodium intake levels-even after
extensive efforts had been made to
identify all possible foods that might
have such an influence.
The checklist is useful but it has
limitations. For example, it is difficult
to code mixed dishes such as pizza and
spaghetti with meat sauce accurately
since individual recipes may vary
greatly in their fat and sodium contents
from one setting to another. Therefore,
individual scores may need to be
adjusted when the checklist is used
with other populations. Portion size
and frequency of consumption were
not specified on the food checklist·
but they may have influenced intakes
of target nutrients reported in 24-hour
recalls. Coders may have been in­accurate
in identifying checklist items
from information on recalls; although
when the checklist is used with other
populations, we believe such errors
were small.
The food checklist we developed was
designed to assess group level differ­ences
by gender or between interven­tion
and control groups, and not
individual intakes. Since this checklist
asks only about !-day's intake, a single
administration cannot be used to assess
habitual dietary intakes of individuals.
There is a large intra-individual
variation in diet, so information from a
single day 's intake-either by 24-hour
recall or by food checklist- is an
efficient way to rank individuals'
habitual nutrient intake. This also can
be used to study the associations
between intakes and physiological or
behavioral risk factors. It is possible
that multiple administrations of the
food checklist would be better indica­tors
of"usual" intakes of the nutrients
studied. However, this hypothesis needs
to be examined and tested further. The
food checklist must be administered to
large samples to obtain the same degree
of precision in detecting differences
in relative intake levels from group to
group that would be achieved using the
24-hour recall.
Food checklists like the one we have
developed are somewhat time- and
population-specific because food
availability and eating habits differ
between groups and over time. Some
groups may have consumed foods not
included in the checklists that were
significant contributors to intakes of
targeted nutrients, or the food supply or
food intake patterns may have changed
over time. Therefore, food checklists,
such as ours, require further testing
and calibration for use with other
populations, and they must be
periodically updated.
The food checklist may be useful as a
supplement to other tools, such as the
Youth Risk Factor Behavior Surveil­lance
System, used in population-based
monitoring systems, in health care, and
in educational settings when the target
group is middle school students and a
brief assessment of dietary intakes of
fat, saturated fat, and sodium is needed
(7) . These and other brief methods for
determining dietary fat levels deserve
Family Economics and Nutrition Review
consideration, keeping in mind issues
of validity for the intended purpose
(31 ,32). A downloadable version of
the checklist, scoring key, and adminis­tration
instructions is available at the
CATCH project Web site, along with
other CATCH data collection forms.
Applications
Techniques described in this article
can be used to develop food checklists
to measure intakes of other nutrients.
The food checklist presented here is
a valid, reliable, and useful tool for
assessing middle school students' food
choices contributing to fat, saturated
fat, and sodium in their diets. A copy
of the checklist and procedures for
administering it are available on the
Internet at http://www.sph.uth. trnc.edu:
8052/chprd/catch/. However, it requires
further testing and calibration before it
can be used with other populations.
Acknowledgments
This research was supported by funds
from the National Heart, Lung, and
Blood Institutes of the National
Institutes of Health (UO 1-HL-39880,
UOl-HL-39852, UOI-HL-39906,
UOl-HL-39927, UOI-HL-39870).
2001 Vol. 13 No.2
References
I. Block, G. 1992. Simplified fat screener. Journal of Nutrition 124:2296S-2298S.
2. Block, G., Clifford, C., Naughton, M., Henderson, M., and McAdams, M. 1989.
A brief dietary screen for high fat intake. Journal of Nutrition Education 21 : 199-
207.
3. Block, G., Dresser, C., Hartman, A., and Carroll, M. 1985. Nutrient sources in
the American diet: Quantitative data from NHANES II Survey I: Vitamins and
minerals. American Journal of Epidemiology 122:13-26.
4. Block, G., Dresser, C., Hartman, A. , and Carroll, M. 1985. Nutrient sources in
the American diet: Quantitative data from NHANES II Survey: Macronutrients and
fat. American Journal of Epidemiology 122:27-40.
5. Block, G., Norris, J., Mandel, R., and DiSogra, C. 1995. Sources of energy
and six nutrients of low income Hispanic-American women and their children:
Quantitative data from NHANES 1982-1984. Journal of the American Dietetic
Association 95 :195-208.
6. Blum, R., Harris, L., Resnick, M., and Rosenwinkel, K. 1989. Technical report
on the Adolescent Health Survey, University ofMinnesota.
7. Byers, T., Serdula, M., Kuester, S., Mendlein, J. , Ballew, C., and McPherson, R.
1997. Dietary surveillance for states and communities. American Journal of
Clinical Nutrition 65 :1210S-1214S.
8. Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences.
Lawrence Erlbaum Associates, Mahwah, NJ.
9. Dennison, B. 1994. Young Children's Diet Assessment Questionnaire. Journal
of Nutrition 124:2303 S.
10. Hampl, J. and Betts, N. 1995. Comparisons of dietary intake and sources offat
in low- and high-fat diets of 18- to 24-year olds. Journal of the American Dietetic
Association 95:893-897.
11. Kann, L. , Warren, W., Collins, J., Ross, J., Collins, B., and Kolbe, L. 1993.
Results from the national school-based 1991 Youth Risk Behavior Survey and
progress toward achieving related health objectives for the nation. Public Health
Reports 108(supp1):47-67.
12. Kelder, S., Perry, C., Klepp, K., and Lytle, L. 1994. Longitudinal tracking of
adolescent smoking, physical activity and food choice behaviors. American
Journal of Public Health 84:1121-1126.
13. Kelder, S., Perry, C., Lytle, L., and Klepp, K. 1995. Community-wide youth
nutrition education: Long term outcomes of the Minnesota Heart Health Program.
Health Education: Research, Theory and Practice 10: 119-131.
9
10
14. Krista!, A., Abrams, B., Thomquist, M., DiSogra, L., Croyle, R., Shattuck, A. ,
and Henry, H. 1990. Development and evaluation of a food use checklist for
evaluation of community nutrition interventions. American Journal of Public
Health 89: 1318-1322.
15. Landis, R. and Koch, G. 1997. The measurement of observed agreement for
categorical data. Biometrics 33:159-174.
16. Luepker, R.V., Perry, C.L., McKinlay, S.M., Nader, P.R., Parcel, G.S., Stone,
E.J. , Webber, L.S., Elder, J.P., Feldman, H.A., Johnson, C.C., Kelder, S.H., and
Wu, H. 1996. Outcomes of a field trial to improve children 's dietary patterns and
physical activity: The Child and Adolescent Trial for Cardiovascular Health
(CATCH). lAMA 275:768-776.
17. Lytle, L., Kelder, S., Perry, C., and Klepp, K. 1995. Covariance of adolescent
health behaviors: The Class of '89 Study. Health Education: Research, Theory
and Practice 10:133-146.
18. Lytle, L.A. , Nichaman, M.Z., Obarzanek, E., Glovsky, E., Montgomery, D.,
Nicklas, T., Zive, M., and Feldman, H. 1993. Validation of 24-hour recalls
assisted by food records in third-grade children. Journal of the American Dietetic
Association 93: 1431-1436.
19. Parcel, G., Edmundson, E., Perry, C., Feldman, H., O'Hara-Thompkins, N.,
Nader, P., Johnson, C., and Stone, E. 1995. Measurement of self-efficacy for diet­related
behaviors among elementary school children. Journal of School Health
65:23-27.
20. Perry, C., Klepp, K., and Sillers, C. 1989. Community-wide strategies for
cardiovascular health: The Minnesota Heart Health Youth Program. Health
Education: Research, Theory and Practice 4:87-101.
21. Perry, C., Stone, E., Parcel, G., Ellison, R., Nader, P., Webber, L., and
Luepker, R. 1990. School-based cardiovascular promotion: The Child and
Adolescent Trial for Cardiovascular Health. Journal of School Health 60:406-
413 .
22. Plan and Operation of the Third National Health and Nutrition Examination
Survey, 1988-94. 1994. Centers for Disease Control and Prevention/National
Center for Health Statistics, Hyattsville, MD.
23. Posner, B., Jette, A., Smith, K., and Miller, D. I 993. Nutrition and health risks
in the elderly: The Nutrition Screening Initiative. American Journal of Public
Health 83:972-978.
24. Resnick, M., Bearman, P., Blum, R., Bauman, K. , Harris, K., Jones, J., Tabor,
J., Beuhring, T. , Sieving, R. , Shew, M., Ireland, M., Beringer, L., and Udry, J.
1997. Protecting adolescents from harm, findings from the National Longitudinal
Study on Adolescent Health. lAMA 278:823-832.
25. Rockett, H., Wolf, A., and Colditz, G. 1995. Development and reproducibility
of a food frequency questionnaire to assess diets of older children and adolescents.
JAMA 95 :336-340.
Family Economics and Nutrition Review
26. Simons-Morton, B., Baranowski, T. , Parcel, G., O' Hara, N. , and Matteson, R.
1990. Children's frequency of consumption of foods high in fat and sodium.
American Journal of Preventive Medicine 6:218-227.
27. Smith, K., Hoelscher, D.M., Lytle, L.A., Dwyer, J.T., Nicklas, T.A., and Zive,
M.M. 200 I. Reliability and validity of the CATCH food checklist: A self report
instrument to measure fat and sodium intake by middle school students. Journal of
the American Dietetic Association I 0 I :635-642; 647.
28. Thompson, F. and Byers, T. 1994. Dietary Assessment Resource Manual.
Journal of Nutrition 124:2245S-2311S.
29. Witschi, J. , Ellison, R., Doane, D., Vorkink, G., Slack, W., and Stare, F. 1985.
Dietary sodium reduction among students: Feasibility and acceptance. Journal of
the American Dietetic Association 85:816-821.
30. Witschi, J.C., Capper, A.L., and Ellison, R.C. 1990. Sources of fat, fatty acids,
and cholesterol in the diets of adolescents. Journal of the American Dietetic
Association 90: 1429-1431 .
31. Yaroch, E.A. 2000. Eight faces of validity. Journal of the American Dietetic
Association I 00(2) :256.
32. Yaroch, A.L., Resnicow, K., and Khan, L.K. 2000. Validity and reliability of
qualitative dietary fat index questionnaires: A review. Journal of the American
Dietetic Association I 00(2) :240-243.
2001 Vol. 13 No.2 II
Gabriela Castellani, MS
Purdue University
Sharon A. DeVaney, PhD
Purdue University
12
Using Credit to Cover Living
Expenses: A Profile of a Potentially
Risky Behavior
Although previous research has examined people's general attitude toward
using credit, no previous research has examined factors that influence people's
attitude toward the use of credit when their income is cut. This study explored
people's attitude toward borrowing money to cover living expenses when
income is cut. The 1995 Survey of Consumer Finances (SCF) was used to
examine attitude toward the use of credit. A multivariate logistic regression
analysis showed that households who were younger, non-White, with less
household income, and who incurred late debt payments were more likely to
say that it was acceptable to use credit to cover living expenses when income
was cut. The findings suggest a need for education targeted to specific groups
of adults and the need for personal finance education for high school students,
the consumers of the future.
T he use of credit is an accepted
practice in the United States.
Households are able to meet
their wants and needs by using various
forms of credit available in the market.
Several factors have been associated
with growth in consumer debt: such as
higher incomes, a general increase in
both the standard and level ofliving,
the marketing of new forms of credit,
and a greater acceptance of debt (20).
The wider distribution of credit cards
could indicate that lenders are including
a larger number of risky borrowers (3)
who are likely to include households
with lower or less stable incomes. If so,
it could be important to study how these
households feel about using credit in a
stressful situation, such as during the
loss or reduction of income.
Research on the use of credit has shown
that attitudes toward credit usually
constitute good predictors of credit use.
Studies in 1970, 1986, 1993, and 1996
have found that attitudes are signifi­cantly
related to the use of credit cards
(6,7,10, 17). Panel data from the 1983
and 1989 Survey of Consumer Finances
(SCF) provide information about the
proportion of households who believe it
is acceptable to borrow to cover living
expenses when income is cut (1 3).
Researchers have shown that consumers
with a positive attitude toward the use
of credit were more likely to use credit
cards from both banks and retail stores
(1 0), and 43 percent of these credit card
users have said it was acceptable to
borrow to cover living expenses (7).
People with favorable attitudes toward
borrowing are more likely not to pay
their monthly credit card balances
in-full at the end of the month, com­pared
with those who do (7) . Other
researchers have shown that consumers
who think it is acceptable to borrow
had a higher credit card balance than
do those with negative attitudes toward
borrowing (4). Further, people who
thought of themselves as "upper class"
believed it was more appropriate to
borrow to purchase luxury goods than
did people of lower or middle socio­economic
status (1 7).
Family Economics and Nutrition Review
Various aspects of financial status and
household demographic characteristics
(e.g., age, marital status, household
size, race, and life cycle stage) have
been examined in previous studies.
Although the focus of the studies, the
sources of data, and the methods differ
slightly, the findings suggest that
specific demographic characteristics
are frequently related to income and
payment difficulties.
Census Bureau data were used to
describe changes in the composition
of American households from 1980 to
1988 (19) . Households headed by a
person younger than 25 had the most
serious financial problems because they
tended to have low incomes and were
likely to face difficulties when meeting
their basic household needs. In a study
using data from the 1990 Survey of
Consumer Attitudes, researchers
found that household heads who
were divorced or separated, had more
children under 18 years of age, and
who had a low level of education had
problems paying their credit obligations
on time (9). Other investigators studied
changes in household debt by using
three cross-sectional studies: the 1983,
1989, and 1992 SCF (8). These
households showed that households
headed by young people and non­Whites
had a high incidence of late
credit payments. Other studies showed
that age was related negatively to the
amount of debt carried by households
(20,21) .
Other factors that might affect the use
of credit when income is cut include
level of education, health status, and
the possibility of receiving government
health insurance. A low level of
education is likely to mean that people
have jobs or occupations with lower
pay and could also mean that people
are less likely to understand the
terminology or information about
lending that is used or made available
2001 Vol. 13 No.2
in the borrowing process (3,5) . A
study comparing borrowers and non­borrowers
found that borrowers spent
more money on health insurance and
prescription drugs and medical equip­ment,
believed to be due to poor health
(1 1).
Another approach to examining
income and payment difficulties is
to consider the household's economic
characteristics. Research has shown
that low-income households have the
highest debt payment-to-income ratio
and few fmancial assets to meet their
payment obligations (8) . Also, a high
percentage of these households have
reported having income levels lower
than they expected, which affected
their ability to pay debts as scheduled.
Further, the households with a high
incidence of late payments tended to
have both low income and little net
worth. In another study, researchers
found that households with payment
difficulties had low incomes and high
debt payment-to-income ratios and
were renters (9) .
A study exploring consumer debt
burden revealed that as net income
and total assets increased, consumer
debt increased, and as consumer debt
increased, year-end savings declined
(20) . A study of credit card use in poor
households suggested that the increased
use of credit by poor families may be
related to a decrease in welfare funding
(2).
No previous research has examined
factors such as demographic and
economic characteristics that might
determine consumers' attitudes toward
borrowing when income is cut. Thus,
the purpose of this exploratory study is
to develop a profile of households who
say they will use credit to cover living
expenses when income is cut and to
examine factors that might explain that
attitude. Using credit as a protection
against the hardship of losing income
resembles the use of precautionary
savings to smooth consumption. Unlike
savings, the use of credit leaves
households with a debt that may be
difficult to pay, especially when
household income is low. A focus on
this problem is relevant for consumer
educators and lenders. The findings of
this study will provide helpful informa­tion
to consumer educators who can
target those households who would
benefit from learning how to manage
their finances more effectively and to
lenders who are likely to learn more
about the households who represent a
higher risk.
Methods
Data and Sample
We used data from the 1995 SCF,
which provides detailed information
on financial and demographic charac­teristics
ofU.S. households and is
sponsored by the Federal Reserve
Board and other agencies (16). The
1995 SCF consists of 4,299 households.
Of these, 2,780 families were selected
by using a standard multistage prob­ability
design. The other I ,519 families
were selected by using a special list
drawn from tax records to oversample
wealthy families. For our study, the
entire sample of 4,299 households was
used and weighted to represent the
population of interest. To deal with
missing information on individual items
in survey data, analysts at the Federal
Reserve Board used multivariate
statistical methods to impute missing
data. Imputation of missing data results
in a multiple number of complete data
sets. Since 1989, the SCF uses multiple
imputation techniques to deal with
missing data. This procedure creates
five data sets (called "implicate" data
sets). ln this study, we use the first
implicate.
13
Variables
The dependent variable was developed
from one of the questions in the 1995
SCF, which was asked by a facilitator,
that measured attitude toward specific
uses of credit: "People have many
different reasons for borrowing money
which they pay back over a period of
time. For each of the questions I read,
please tell me whether you feel it is
all right for someone like yourself to
borrow money." The choices were "to
cover living expenses when income is
cut, to cover the expenses of a vacation
trip, to fmance the purchase of a fur
coat or jewelry, to fmance the purchase
of a car, or to fmance educational
expenses." Each part of the question
was answered with a "yes" or "no."
Only the question "to cover living
expenses when income is cut" was
selected for study. The dependent
variable was "Is it all right to borrow
money when income is cut?" It was
coded as I if the response was "yes"
and 0 for "no" (table 1). To examine
the relationship between this dichoto­mous
dependent variable and the
independent variables, we used a
logistic regression (1 5).
The independent variables represent
demographic, economic, credit, and
attitudinal factors. The demographjc
variables consisted of age, marital
status, race, education, and household
size. Age was coded as a categorical
variable with four groups: household
heads younger than 35 years old, 35 to
44, 45 to 54, and 55 or older. These
categories were intended to represent
the life cycle stages of the household
(16,20).
Race was coded as 1 if the household
head was White and 0 otherwise;
marital status was coded as I if the
household head was married and 0 if
otherwise (1 6) . The highest level of
education attained by the household
head and household size were
continuous variables.
14
Table 1. Coding of dependent and independent variables
Variable
Dependent
Do you feel it is all right to borrow money to
cover living expenses when income is cut?
Independent
Age
Less than 35
35-44
45- 54
55 and older (reference group)
Marital status
Race
Level of education
Household size
Household income
Less than $10,000
$10,000- $19,999
$20,000- $29,999
$30,000- $49,999
$50,000 or more (reference group)
Home ownership
Liquid assets
Government health insurance
Number of credit cords
Payment pattern
No payment obligations (reference group)
Late payments
Payment on schedule
Credit cord balance outstanding
Expectation about income
Self-reported health
1Separated, divorced, widowed, and never married.
Measurement
1 =yes, 0 = no
=yes, 0 = no
=yes, 0 = no
=yes, 0 = no
= yes, 0 = no
= married, 0 = otherwise 1
= White, 0 = otherwise2
Continuous
Continuous
= yes, 0 =no
= yes, 0 = no
= yes, 0 = no
=yes, 0 = no
=yes, 0 =no
= renter, 0 = homeowner
Continuous
1 = eligible, 0 = otherwise
Continuous
=yes, 0 =no
= yes, 0 = no
= yes, 0 = no
Continuous
1 = income is lower than
expected, 0 = no
1 = health is fair or poor,
0 = otherwise
2Biack or African American, Hispanic, Asian or Pacific Islander, Native American, and Other.
The economic variables included total
annual household income, home­ownership,
amount ofliquid assets,
and eligibility for government he~lth
insurance. Income was coded as a
categorical variable. Amount ofliquid
assets was used as a continuous variable
and was calculated by summing the
amount of money in savings, checking,
money market deposit accounts, and
call accounts at brokerages. Renter
was coded as 1, and homeownership
was coded as 0. Government health
insurance was coded as 1 if the reply
to the following question was positive:
"Are you or anyone in your family
living here, including household
members with independent finances,
currently eligible to receive benefits
from any government health insurance
programs, such as Medicare, Medicaid,
or CHAMPUS, VA (Veterans' Assis­tance),
or other military programs?"
We included government health
insurance because the receipt of this
benefit could be a resource for house­holds
when income was cut (2).
Family Economics and Nutrition Review
Figure 1. Distribution of households answering: "Do you feel it is all right to
borrow money to cover living expenses when income is cut?"
The credit-related variables included
number of credit cards, payment
pattern, and outstanding balance on
credit card after the last monthly
payments were made. Number of credit
cards, coded as a continuous variable,
was used as a proxy for experience in
using credit. The outstanding balance
on credit cards was treated as a
continuous variable. Payment pattern
was measured by the response to the
question, "Now thinking of all the
various loan or mortgage payments
you made during the last year, were all
the payments made the way they were
scheduled, or were payments of any
of the loans sometimes made later or
missed?" The responses were "always
pay debt as scheduled, sometimes got
behind or missed payments, and
inapplicable." The households for
whom the question was "inapplicable"
were identified as having no payment
obligations and were therefore used as
the reference group.
The attitudinal variables included the
household head's perception of their
income for the last year and his or her
personal health status. Perception of
income measured how the level of
income was viewed in relation to what
2001 Vol. 13 No.2
was expected in a normal year. This
variable was coded as 1 if income was
lower than expected and 0 if otherwise.
Health status was coded as 1 if the
household heads reported their health
status as fair or poor and 0 if otherwise.
Results
Description of Sample
Slightly less than half ( 44 percent) of
the household heads said it was "all
right" to borrow money to cover living
expenses when income was cut (fig. 1).
The average household size was two
people, and the household head had
completed almost 13 years of education
(table 2). One-fourth of the households
were headed by a person younger than
35; three-fourths, by a person who
was White; and a little over half, by
a person who was married. Sixteen
percent of the households had annual
household income below $1 0,000;
50 percent had household incomes
of $30,000 or more. Over half were
homeowners: 57 percent. Slightly more
than one-third of the households were
eligible for some type of government
health insurance: 38 percent.
Households whose heads
are younger, non-White, with
household income below
$20,000, and who had
incurred late debt payments
are more likely to borrow
money-use credit-to cover
living expenses when income
is cut.
15
Table 2. Description of households, 1995 Survey of Consumer Finonces1
Variable
Household size
Years of education
Liquid assets
Number of credit cords
Credit cord balance outsta nding
Age
Less than 35
35 - 44
45-54
55 and older
Marital status
Married
Not married
Race
White
Non-White
Household Income
Less than $10,000
$10,000- $19,999
$20,000- $29,999
$30,000- $49,999
$50,000 or more
Homeownership
Homeowners
Renters
Government health insu rance
Eligible
Non-eligible
Payment pattern
No payment obligations
Late payments
Payment on schedule
Expectation about income
Income lower than expected
Income as high or higher than expected
Health status
Fair or poor
Very good or excellent
' N=4,299.
16
Meas urement
Mean
(Median)
2.38
(2)
12.9
(12)
$13,258
($1,600)
1.61
(1)
$1,647
($424)
Percent
24.8
23.0
17.9
34.4
52.5
47.5
77.6
22.4
16.4
18.6
14.6
24.0
26.0
56.7
43.3
37.7
62.3
35.3
16.5
48.2
16.4
83.6
24.5
75.5
Whereas the average amount of I iquid
assets was $13 ,258, the median was
only $1 ,600. The average amount of
outstanding credit card balance was
$1 ,647, while the median balance was
considerably lower: $424. On average,
households held one to two credit
cards. Almost half ( 48 percent) of the
households in the sample reported that
they paid their debts on schedule while
17 percent reported being late or
missing payment obligations. Thirty­five
percent had no payment obliga­tions.
One-fourth of the household
heads perceived their health status
as fair or poor, and over four-fifths
reported that their income had been as
high or higher than what they expected
for a normal year, 25 and 84 percent,
respectively.
Predictors of Attitude Toward
Use of Credit
The factors that were statistically
significant predictors of having a
positive attitude toward using credit
when income was cut were age, income,
being a non-minority, and payment
pattern (table 3). The odds that the
head of household will borrow to cover
living expenses when income is cut
increase from 46 to 94 percent for
household heads younger than 35
(94 percent), those aged 35 to 44
(57 percent), and 45 to 54 ( 46 percent),
compared with households headed by
a person age 55 and over. When the
head of household is White, the odds
that the head will borrow to cover
living expenses when income is cut
decrease by 16 percent, compared
with a non-White head of household.
The odds that households wiJI borrow
when income is cut increased signifi­cantly
for those with incomes less than
$10,000 and between $ 10,000 and
$19,999, compared with households
with more than $50,000 yearly income.
The odds that a household with an
income less than $1 0,000 would borrow
money when income was cut increased
Family Economics and Nutrition Review
Table 3. Results of logistic regression: Attitude toward borrowing when income is cut 1995 Surve of C
Financesl , y onsumer
Variab le Parameter estimate P-volue Odds ratio
Age (55+ reference group}
Less than 35 .6608 .0001*** 1.936
35 - 44 .4512 .0001 *** 1.570
45 - 54 .3795 .0001 *** 1.462
Married -.0866 .2540 0.91 7
White -.1698 .0481 * 0.844
Education .0115 .3858 1.012
Household size .0134 .5843 1.013
Household income ($50,000+ reference group)
Less than $1 0,000 .3890 .0045** 1.475
$10,000-$ 19,999 .2928 .0159* 1.340
$20,000 - $29,999 .0920 .4419 1.096
$30,000- $49,999 .0606 .5110 1.063
Renter - .0879 .2690 0.916
Liquid assets -2.41 E-8 .3599 1.000
Eligible for government health insurance .0464 .5588 1.047
Number of credit cords - .00421 .8239 0.996
Payment pattern (no payment obligation, reference group)
Payment on schedule - .0128 .8728 0.987
Late payment .2725 .0214* 1.288
Credit card balance .000013 .0829 1.000
Income lower than expected - .0495 .6380 0.952
Poor health -.0253 .7693 0.975
Intercept - .6166 .0132*
-2 LOG likelihood 5,7 43.488***
1N= 4,299.
*P<.OS; **P <.Ol ; ***P<.OOl .
by 48 percent, compared with the
household that had a $50,000 income.
The household with income between
$10,000 and $19,999 increased its
odds of borrowing money by 34
percent. When the household is late
with payments, the odds increase by 29
percent that the household will borrow
money to cover living expenses when
income is cut, compared with house­holds
with no payment obligations.
Discussion and
Implications
Households whose heads are younger,
non-White, with household income
below $20,000, and who had incurred
2001 Vol. 13 No. 2
late debt payments are more likely to
borrow money- use credit- to cover
living expenses when income is cut.
These findings support previous studies
on general credit use.
Several findings from other studies,
however, were not supported in the
study. Marital status, liquid assets,
level of education, household size,
homeownership, eligibility for govern­ment
health insurance benefits, number
of credit cards, and health status were
not related significantly to using credit
to cover living expenses when income
is cut. Although the relationship
between outstanding credit card balance
and the dependent variable was not
significant, it was positive. This
suggests that consumers with larger
balances would charge more if their
income was cut.
This study provides information about
consumers who consider it appropriate
to use credit when there are income
difficulties. These households appear to
be more likely to use credit when they
face unemployment or unexpected
events such as illness or accidents that
affect the level of their household
income. A previous study has pointed
out that there are different types of
borrowers, such as some who borrow
for the purpose of social display and
others who borrow to cover expendi­tures
on necessities (I 1). It may be
difficult to reach younger, low-income
households that are having difficulty
paying on time through educational
17
programs. A type of educational
program that is gaining more attention
is Personal Finance Employee Educa­tion
at work (12) . The potentially risky
households who were identified here
are likely to benefit from education
provided at the workplace that would
help them understand the potential
consequences of not paying off debts,
finding strategies to reduce debt load,
or identifying community and govern­ment
resources that increase income
or reduce expenses. Also, education
provided by the Cooperative Extension
Service, faith organizations, and other
groups would be beneficial (1).
Another technique for helping
consumers manage money better is to
support the continued implementation
of the NEFEg High School Financial
Planning Program (14). If high school
students learn about budgeting and
using credit, the knowledge and skills
gained while they are students may be
more likely to continue as they enter
college and the work force. Another
alternative available to consumers is
the Neighborhood Financial Care
Center (formerly known as Consumer
Credit Counseling Services). The
Center helps consumers evaluate
and pay down their debt.
The finding that having difficulty
making payments on time increases
the likelihood of borrowing when
income is cut is a complex issue.
Lenders may have extended credit to
people who had good credit histories
but who are now having difficulties
(because of unemployment or health
problems, etc.) repaying their debts.
Also, some lenders may have extended
credit to more risky consumers, because
the lender wanted to increase its
customer base. It may be impossible
for consumer educators to address this
issue, but at the local level, consumer
educators can communicate their
concerns to business leaders. The
findings of this study would also be
18
helpful for credit card issuers. Young,
low-income, non-White, and "late
payment" households constitute an
especially high-risk consumer because
they consider it appropriate to use
credit when income is cut, and they
may have few economic resources and
be employed in less stable jobs (3).
Borrowing to cover living expenses
when income is cut should be re­examined
in other ways by using
information that is not available in the
SCF. Work status might be an impor­tant
predictor of attitudes toward
borrowing. Those who are unemployed
temporarily, or those who are employed
in cyclical occupations, may be more
likely to use credit to cover living
expenses when income is cut (18).
Thus it may be necessary to use data
on employment status to understand
better which households will encounter
this problem. Future attempts to answer
the question about the use of credit
when income is cut will surely benefit
consumers who are most in need of this
help.
Family Economics and Nutrition Review
References
I. Adamson, C.R., Mayer, R.G., and Williams, F.L. 1999. Hispanic financial
counseling. Proceedings of the Association for Financial Counseling and
Planning Education, p. I 02.
2. Bird, E.J ., Hagstrom, P.A., and Wild, R. 1997. Credit Cards and the Poor.
Institute for Research on Poverty. Discussion Paper 1148-97.
3. Black, S.E. and Morgan, D.P. I 999. Meet the new borrowers. Federal Reserve
Bank of New York Current Issues in Economics and Finance 5(3): 1-6.
4. Bloom, D.E. and Steen, T.P. I 987. Living on credit. American Demographics,
October, pp. 22-29.
5. Brobeck, S. 1992. Consumers' attitudes toward credit cards. Credit World, July/
August, pp. 8-13.
6. Calem, P.S. and Mester, L.J. 1993. Search, switching costs and the stickiness of
credit card interest rates. Working paper number 92- 24/R. Federal Reserve Bank
of Philadelphia, Philadelphia, PA.
7. Canner, G.B. and Cymak, A.W. 1986. Determinants of consumer credit card
usage patterns among U.S. families. Journal of Retail Banking VII(3):63-74.
8. Canner, G.B. and Kennickell, A.B. 1995. Household sector borrowing and the
burden of debt. Federal Reserve Bulletin 81:323-334.
9. Canner, G.B. and Luckett, C.A. 1991 . Payment of household debts. Federal
Reserve Bulletin 77:218-229.
I 0. Choi, H.N. and DeVaney, S.A. 1996. Differences in the use of bank and retail
credit cards in the U.S.A. Journal of Consumer Studies and Home Economics
19:381-392.
II . Fan, J .X. 2000. Linking consumer debt and consumer expenditures: Do
borrowers spend money differently? Family and Consumer Sciences Research
Journal 28(3):357-400.
12. Garman, E. T. 1999. Employer-sponsored education programs and incentives
to improve employees' financial lifestyles. Personal Finances and Worker
Productivity 3(2):3-8.
13. Godwin, D.D. 1997. Dynamics of households ' income, debt and attitudes
toward credit, 1983-1989. The Journal of Consumer Affairs 31 (2):303-325 .
14. Huddleston-Casas, C.A., Danes, S.M., and Boyce, L. 1999. Impact evaluation
of a financia11iteracy program: Evidence for needed policy change. Consumer
Interests Annual45: 109-114.
2001 Vol. 13No.2 19
15 . Kennedy, P. 1998. A Guide to Econometrics (4th ed.). TJ International,
United Kingdom.
16. Kennickell, A.B., Starr-McCiuer, M., and Sunden, A.E. 1997. Family finances
in the U.S. : Recent evidence from the Survey of Consumer Finances. Federal
Reserve Bulletin 83(1): 1-24.
17. Slocum, J.W. and Mathews, H.L. 1970. Social class and income as indicators
of consumer credit behavior. Journal of Marketing 34:69-74.
18. Sullivan, T.A., Warren, E. , and Westbrook, J.L. 2000. The Fragile Middle
Class: Americans in Debt. Yale University Press, New Haven, CT.
19. Waldrop, J. 1989. Inside America's households. American Demographics,
March, pp. 20-27.
20. Wasberg, C.A., Hira, T.K., and Fanslow, A.M. 1992. Credit card usage and
consumer debt burden of households. Journal of Consumer Studies and Home
Economics 16:9-32.
21. Zhu, L.Y. and Meeks, C.B. 1994. Effects of low income families' ability and
willingness to use consumer credit on subsequent outstanding credit balances.
The Journal of Consumer Affairs 28(2):403-422.
20 Family Economics and Nutrition Review
Hazel A.B. Hiza, PhD, RD, LN
USDA, Center for Nutrition Policy
and Promotion
Allan A. Johnson, PhD, LN
Howard Univers ity
Enid M. Knight, PhD, LN
Howard University
Claudette S. Welch, PhD, LN
Bloomberg School of Public Health
Johns Hopkins University
Cecile H. Edwards, PhD, LN
Howard University
2001 Vol. 13 No.2
Relationships of Substance Abuse
to the Nutritional Status of
Pregnant African-American
Women
The effect of illicit drug use, which was determined from fasting blood samples,
on maternal nutritional status was examined in a study of African-American
pregnant women. Participants were classified as drug users, trace drug users,
and nondrug users. Quantitative self-reported dietary records and maternal
anthropometric measurements were collected . Consumption of protein, vitamin
A, ascorbic acid, selected B-complex vitamins, and phosphorus equaled or
exceeded 1 00 percent of the 1 989 Recommended Dietary Allowances (RDA)
for all groups. Vitamin B6
, calcium, folate, iron, magnesium, and zinc were
consumed in amounts below 1 00 percent of the 1 989 RDA. Food energy,
nutrient intakes, sociodemographic characteristics, maternal anthropometric
measurements, and delivery weight were similar among the three groups. The
unexpected results of this study may be due to the method used to classify the
participants. Thus, more extensive research is needed .
I llicit and nonillicit drug abuse
is a major interest of clinicians,
public health officials, and
social authorities (e.g., child welfare).
Moreover, one of the major concerns is
drug abuse during the periconceptional
period and throughout pregnancy
because of its potential adverse effects
on the health of the mother, embryo,
fetus, and neonate (14,36,37) .
Age, race, and socioeconomic status
are among the most frequently cited
factors associated with low birth weight
and preterm delivery. Specifically,
being young, being African American,
and having a low socioeconomic status
are most often associated with adverse
pregnancy outcomes (1,1 J,18,19,32,
39). A higher maternal educational
level is associated with better health
knowledge and behavior (35).
One to 58 percent of pregnant women
use drugs (47). Such wide variations in
reported use could be attributed to the
voluntary nature and lack of adequate
drug-screening techniques, disparate
patterns of drug use among different
U.S. regions and populations, differ­ences
in drug-screening methods, or
differences in levels of prenatal care
among drug-using populations (27) .
Lack of agreement exists in the
scientific literature regarding the most
prevalent illicit drugs used during
pregnancy. However, research shows
that about II percent of pregnant
women in the United States use at least
one of the following drugs: cocaine,
marijuana, heroin, methadone, phencyc­lidine
(PCP), and amphetamines (40).
Each year in this country, more than
200,000 infants are exposed in utero to
one or more illicit drugs (9,45) .
21
Women who abuse illicit drugs and
alcohol during pregnancy are an elusive
population. These women often remain
unidentified to practitioners and
researchers and therefore have not been
studied to a great extent (22). Despite
the even distribution of illicit substance
use across demographic categories,
poor women and women of color are
far more likely than are other women to
be reported to health and child welfare
authorities for use of substances during
pregnancy, even when their base rates
for use of illicit drugs are considered
(22) .
Little information is available on the
nutritional consequences of substance
abuse during pregnancy, and the
available studies of women who have
used nonillicit as well as illicit drugs
during pregnancy have provided
conflicting results regarding the
nutritional effect on users (26,30) .
Some evidence shows that cocaine
acts as an appetite suppressant (52) .
Another shows increased caloric intake
and low levels of plasma zinc among
marijuana users (29) . Researchers
estimate that nearly 50 percent of
opiate-dependent women suffered
from anemia, heart disease, diabetes,
pneumonia, or hepatitis during
pregnancy and childbirth (52) .
Another study shows that women who
consumed alcohol during pregnancy
drank more frequently before preg­nancy
than did women who drank
alcohol prenatally but not during
pregnancy (33). Jacobson and others
(25) also found that many mothers
reported higher levels of alcohol
consumption before pregnancy than
during pregnancy. One plausible
interpretation is that the mothers
underreported their actual levels of
drinking when they were interviewed
at prenatal clinics because of the
stigma associated with drinking during
pregnancy. This may be especially
likely when women are interviewed
22
in a prenatal clinic where the health
and welfare of the infant is focal.
Alternatively, self-reported alcohol
consumption by pregnant women may
be influenced by their current level of
drinking, which is typically higher.
Excessive alcohol consumption
impairs the metabolism of most
nutrients. Ethanol intake also leads
to negative nitrogen balance and an
increased protein turnover (8,52).
However, evidence concerning the
adverse effects of alcohol on specific
nutritional indices comes mainly from
studies of nonpregnant, hospitalized
alcoholics; few data are available on the
effect of alcohol on maternal nutrition
(52). Information is particularly sparse
on the diets of pregnant women of
African descent and almost nonexistent
for pregnant women who are substance
abusers. In one study, maternal and
umbilical cord blood zinc levels were
lower in pregnant women who con­sumed
alcohol than in those who did
not (16). Another study suggested
that alcohol may impair placental
transport of amino acids (I 5).
Another behavior--cigarette smoking­may
affect maternal nutrition by de­creasing
the availability of calories
and certain nutrients such as vitamin
B12
, amino acids, folate, and zinc (52).
Efforts to improve maternal and fetal
nutrition during pregnancy have
focused on achieving appropriate
energy intakes and ensuring that the
intake of specific nutrients is adequate
to meet maternal and fetal requirements
(52).
Despite researchers' efforts in recent
years to document the consequence of
maternal substance abuse on pregnancy
outcomes, information on specific
maternal consequences of substance
abuse during pregnancy is sparse. Thus
this study focused on the relationships
ofnonillicit (alcohol and tobacco) and
illicit (cocaine, marijuana, heroin, PCP,
and opiates) substance abuse to the
nutritional status of pregnant African­American
women residing in an urban
environment.
Methods
Research Design and Study
Participants
A prospective research design was
used in the study. Participants were
recruited prior to the twenty-eighth
week of gestation and followed until
the birth of their child. The study par­ticipants
were 163 African-American
pregnant women who were ages 16 to
35 and had no previous pregnancies
that continued beyond 28 weeks.
Subjects were free of diabetes mellitus
and abnormal hemoglobins (sickle cell
disease, thalassemia, and hemoglobin
C). They were recruited from prenatal
clinics operated by two urban hospitals
and the Department of Health and
Human Services (DHHS).
Data Collection
On entry into the study, participants
were interviewed by trained personnel
who collected sociodemographic data
(age, marital status, educational level
attained, and annual household in­come).
Quantitative dietary data were
collected monthly by using the 24-hour
dietary recall method. Participants
were recruited at various stages of their
pregnancy; thus, the number of recalls
varied from I to 7 days, with a mean of
2.6 days. We used three-dimensional
food models and various measuring
implements (measuring cups, spoons,
etc.) to help participants recall how
much foods and beverages were
consumed the previous day. The
Nutriplanner 6,000 System was used
to calculate food and nutrient intake
data (42).
The use of illicit and nonillicit drugs
was determined by self-reports and
Family Economics and Nutrition Review
biochemical analyses. After recruitment
into the study, the women were asked
whether they bad used alcohol, cocaine,
marijuana, heroin, opium, or PCP
before and during pregnancy. Fasting
venous blood samples were collected
from the participants during each
trimester: 1-13 weeks, 14-26 weeks,
and 27 or more weeks. The prevalence
of self-reported drug use before and
during pregnancy was compared with
the biochemical determination of drug
use. Weeks of gestation at birth were
established (10) .
Analyses for cocaine, marijuana,
opium, or PCP were conducted on
aliquots of serum collected from clotted
blood samples that bad been stored
at -80°C. Participants' anthropometric
measurements- pre-pregnancy weight
(self-reported), maternal height,
pregnancy weight gain, and delivery
weight (based on measurements )-were
obtained from their medical records.
The initial semiquantitative testing of
serum samples for illicit drug abuse
was conducted by using the immune
technique that is direct, automated,
and enzyme-mediated (48) .
The classification of participants as
drug users, trace drug users, or nondrug
users was derived by using standards
established by the Alcohol, Drug
Abuse, and Mental Health Administra­tion/
National institutes of Health
Administration on Drug Abuse (I 3) .
Women were classified as drug users
(n= 19) when their serum threshold
levels were at least 300 ng/ml for
cocaine, I 00 ng/ml for marijuana,
300 ng/ml for opiates, or 25 ng/ml for
PCP. Women were classified as trace
drug users (n= l22) when their sera
tested positive for cocaine, marijuana,
opiates, or PCP, but concentration
levels were below the serum threshold
levels for this group. Participants were
classified as nondrug users (n=22)
when their sera showed no evidence
of cocaine, marijuana, opiates, or PCP.
2001 Vol. 13 No, 2
Statistical Methods
Chi-square tests were used to compare
sociodemographic characteristics,
patterns of drug usage, and dietary
practices of pregnant African-American
women who were drug users, trace drug
users, or nondrug users. Analysis of
variance (ANOVA) and Duncan's
multiple range t tests were used to
investigate the relationships of sub­stance
abuse to dietary intakes and
anthropometric measurements among
the three groups of women. The
computer Statistical Package for the
Social Sciences (SPSSx) was used to
analyze the data (50) .
Results
Sociodemographic
Characteristics and Self­Reported
Drug Use
The pregnant African-American women
were ages 16 to 35; most in each group
were age 21 or younger: 58 to 68
percent (table 1). Most of the pregnant
women were single (86 to 95 percent)
and had at least a high school education
(63 to 77 percent). Thirty-two to 53
percent of the women had an annual
household income that was less than
$23,000.
More than 25 percent of the pregnant
women reported using illicit drugs
before pregnancy; this number was
more than eight times greater than the
percentage of pregnant women report­ing
drug use during pregnancy (table 2).
The most commonly abused drug
reported both before and during
pregnancy was marijuana, followed
by cocaine. When interviewed, almost
97 percent of the pregnant women
denied using drugs during pregnancy.
However, biochemical determination
of drug use showed that 88 percent of
the pregnant women were classified as
drug users or trace users.
The most commonly abused
drug reported both before
and during pregnancy was
marijuana, followed
by cocaine.
23
Table 1 . Sociodemographic characteristics of pregnant African-American
women
Group 1
Participants (number)
Age groups (years)
16-18
19-21
22-24
25-27
28-35
Marital status
Single
Married
Other1
Not reported
Highest level of education attained
Elementary school
Some high school
High school graduate
T rode school
College2
Other
Not reported
Annuol household income
< $11,000
$11,000-$22,999
$23,000- $34,999
~ $35,000
Not reported
1Separated, divorced, or cohabitating.
20ne semester or more of college credits.
drug
users
19
21.1
36.8
21 .1
15.7
5.3
94.7
0
5.3
0
0
26.3
47.4
0
15.7
5.3
5.3
15.7
31.6
15.8
5.3
31.6
Group 2 Group 3
trace drug nondrug
users users
122 22
Percent
21 .3 22.7
41.8 45.6
18.0 13.6
10.7 4.5
8.2 13.6
87.0 86.4
9.8 13.6
1.6 0
1.6 0
0.8 0
27.1 22.7
38.5 45.5
5.7 9.1
18.9 22.7
0.8 0
8.2 0
29.5 13.6
23.0 18.2
9.0 13.6
7.4 18.2
31.1 36.4
Sociodemographic characteristics among the three groups were not significantly different (p > O.OS).
Note: Biochemical assays were used to classify the three groups.
24
Among the two nonillicit drugs studied,
cigarettes, compared with alcohol,
were more likely to be used. Eighteen
percent of pregnant women reported
smoking cigarettes during pregnancy;
most smoked 1 to 5 cigarettes per day.
Four percent of the participants
reported consuming alcohol during
pregnancy, with regular beer being the
most popular alcoholic beverage
consumed (fig. 1). Chi-square analysis
revealed no significant relationship
between drug use and smoking or
between drug use and consumption
of alcoholic beverages. However, chi­square
analysis did show a significant
relationship between smoking and the
use of alcoholic beverages (p< 0.05).
This finding indicated that those who
smoked were more likely to use
alcoholic beverages (data not shown).
Energy and Nutrient Intakes
Compared With
Recommended Levels
ln contrast to a priori expectations,
we found that the women who were
classified as drug users had a mean
energy intake that exceeded I 00
percent of the 1989 recommended
energy allowances: I OJ. I percent (41)
(table 3). The other groups of women
had total kilocalorie intakes of Jess than
I 00 percent of these recommendations:
91 to 94 percent. The three groups of
pregnant women had mean intakes of
protein, ascorbic acid, thiamin, ribo­flavin,
niacin, vitamin B12, and phos­phorus
that met or exceeded I 00
percent of the RDAs (table 4). For
ascorbic acid and vitamin B12, the
intakes exceeded 200 percent of the
RDAs: 211 to 259 percent. On the
other hand, intakes of vitamin B6,
folate, calcium, iron, magnesium, and
zinc were less than I 00 percent of the
RDAs: 26 to 82 percent. Drug users
and trace drug users had mean intakes
that exceeded I 00 percent of the RDAs
for vitamin A ( 127 to 151 percent), but
Family Economics and Nutrition Review
Table 2. Prevalence of self-reported drug use of African-American women
before and during pregnancy
Group 1 Group 2 Group 3
Self-reported drug trace drug nondrug
drug use users users users
Participants (number) 19 122 22
Percent
Before pregnancy
Marijuana 15.8 13.1 27.3
Cocaine, heroin, or PCP 10.6 13.0 0
During pregnancy
Marijuana 0 1.6 4.5
Cocaine, heroin, or PCP 0 1.6 0
Self-reported drug use among the three groups was not significantly different (p > 0.05).
Note: Some individuals used more than one drug; therefore, percentages do not total 1 00.
Figure 1 . Use of selected nonillicit drugs by African-American women during
pregnancy
0 5
1 Most smoked 1 to 5 cigarettes per day.
10
Percentage of users
2Beer was the most popular alcoholic beverage.
2001 Vol. 13 No.2
15 20
nondrug users had vitamin A intakes
below I 00 percent of the RDAs
(84 percent). However, the adequacy
offood energy and nutrient intakes
among the groups was not statistically
significant.
Anthropometric
Measurements
The anthropometric measurements
were similar among the three groups
of pregnant African-American women
(table 5). For most of the measure­ments-
pre-pregnancy weight, percent­age
of ideal pre-pregnancy body
weight, body mass index (BMI), and
delivery weight- the means were
highest for drug users, compared with
trace drug users and nondrug users.
The differences, however, were not
statistically significant.
Discussion and
Conclusion
When the sociodemographic character­istics
among three groups of pregnant
African-American women were
compared, no significant differences
were noted. These findings were
comparable to those reported in other
studies that focused on the epidemiol­ogy
of illicit substance abuse and
non illicit drug use. Similar studies
depicted sociodemographic data that
both confirmed (17,33,36,51) and
contradicted (2,46) the fmdings in
this study.
Our study showed that marijuana
was the predominant drug of abuse,
followed by cocaine. The pattern of
illicit self-reported drug use in our
study was similar to the self-reported
pattern of drug use reported by others
who found that marijuana and cocaine
were more likely to be used, compared
25
with opiates (4, 17,25). It was not
surprising that the prevalence of drug
use that is based on self-reports was
lower than the prevalence that is based
on biochemical assays. The low
prevalence of marijuana and cocaine
use reported in this study may be due
to the stigma associated with drug use,
especially during pregnancy, as well as
due to the fear of prosecution. When
the participants in our study were
interviewed, they reported a higher
prevalence of substance abuse before
pregnancy. This finding, which is
confirmed by biochemical determina­tion,
is consistent with results of similar
studies that showed women had been
underreporting their use of illicit drugs
when the interviews occurred during
their pregnancies (23,25). However,
women may be more willing to disclose
retrospectively information regarding
illicit drug use during pregnancy when
it is less likely they will be referred for
treatment, threatened with loss of
custody of their babies, or prosecuted
(23). Further, although self-reported
data are often described as being
inherently unreliable, the accuracy
of self-reports vary considerably
depending on the substance, time of
the interview, skill of the interviewer,
and other factors (25).
One participant, determined by bio­chemical
assays to be a nondrug user,
admitted to being a current drug user.
It is unlikely that a person would admit
to being a current drug user when she is
a nondrug user. Thus it is possible there
is a flaw in the biochemical determina­tion
used in this study to determine
current drug use. Current immunoassay
methods and their routine threshold
levels may not be sensitive enough to
detect serum cocaine, marijuana,
heroin, or PCP in pregnant women.
Also, someone who tested negative for
serum illicit drugs on a given day may
be a heavy drug user who may have
abstained from substance abuse for
26
Table 3. Energy intakes of pregnant African-American women, compared
with the 1989 recommended energy allowances
Group 1 Group 2 Group 3
drug trace drug nondrug
users users users
Participants (number) 19 122 22
Total energy intake (kcal) 2527.0 ± 170.9 2347.0 ± 70.9 2270.8 ± 132.5
1989 Recommended
Energy Allowances (%) 101.1 93.9
Energy intakes among the three groups were not significantly different (p > 0.05).
Note: Biochemical assays were used to classify the three groups.
90.8
Table 4. Nutrient intakes of pregnant African-American women, as
percentages of the 1989 Recommended Dietary Allowances
Group 1 Group 2
drug trace drug
users users
Participants (number) 19 122
Percent RDA
Protein 176.7 160.7
Vitamin A 150.6 126.6
Ascorbic acid 238.2 210.6
Thiamin 126.7 115.8
Riboflavin 144.1 139.6
Niacin 139.0 135.2
Vitamin B6 71 .6 80.5
Folate 58.4 63.3
Vitamin B12 259.0 254.1
Calcium 40.2 32.5
Phosphorus 131 .5 112.0
Iron 50.8 52.5
Magnesium 81.7 75.2
Zinc 76.8 69.2
Nutrient intakes among the three groups were not significantly different (p> 0.05).
Note: Biochemical assays were used to classify the three groups.
Group 3
nondrug
users
22
163.3
84.3
226.3
99.5
118.4
127.2
60.5
46.5
251.8
25.5
103.7
44.9
62.8
60.7
Family Economics and Nutrition Review
Table 5. Maternal anthropometric measurements of pregnant African­American
women
Group 1 Group 2 Group 3
Anthropometric drug trace drug nondrug
measurements users users users
Participants (number) 19 122 22
Mean ± standard error
Height (in .) 63.8 ± 0.6 64.4 ± 0.3 63.5 ± 0.6
Pre-pregnancy weight (lbs.) 144.3 ± 8.7 139.5 ± 3.0 138.4 ± 5.4
Ideal body weight (%) 120.3 ± 6.9 114 .2 ± 2.3 117.4 ± 5.4
Body mass index (BMI) (kg/m2) 24 .9 ± 1.4 23.7 ± 0.5 24.4 ± 1.2
Weekly weight gain (lbs .) 0.7 ± 0.1 0.7 ± 0.04 0.6 ± 0.08
Total weight gain (lbs .) 30.0 ± 4.6 31.1 ± 2.4 24.7 ± 2.8
Delivery weight (lbs .) 173.0 ± 9.0 170.7 ± 3.8 164.4 ± 7.1
Maternal anthropometric measurements among the th ree groups were not significantly different
(p>O.OS).
Note: Biochemical assays were used to classify the three groups.
several days preceding the drug test.
Thus a negative drug test will be read.
In addition, lack of agreement between
self-reports and biochemical determina­tion
of illicit drug use could be partly
due to the relatively short half-life of
most of these illicit drugs. The half-life
of cocaine in the plasma after oral
ingestion or inhalation is I hour. For
marijuana, plasma concentration peaks
within 7 to I 0 minutes; physiological
effects are shown between 20 and 30
minutes. The half-life of PCP appears
to be about 3 days, but it could be
shortened to I day by gastric suction
and acidation of urine (2 1).
Our finding that participants who
smoked cigarettes were more likely to
consume alcoholic beverages, com­pared
with those who did not smoke, is
consistent with results of similar studies
{3,20,25,56) . Other studies showed that
women who drank alcohol during
pregnancy were more likely to smoke
2001 Vol. 13 No. 2
cigarettes and use illicit drugs, to have
parents who drank alcohol, or to feel
that other pregnant women drank
similar amounts of alcohol (2 5) .
The energy intakes of participants in
our study, as a percentage of the
recommended energy allowances (41) ,
were higher than those recorded by
other investigators (7, 1 2,44). In a
similar study, researchers found that
women reporting drug use before
pregnancy had significantly higher
intakes of food energy than did their
counterparts who were using drugs
during pregnancy (2 7). The protein
intakes of the participants in our study
exceeded 161 percent of the RDA and
are consistent with those of other
studies (7, 49). Another study, however,
reported protein intakes of less than
I 00 percent of the RDA for the
pregnant participants who used
illicit drugs (1 2).
Among the two nonillicit drugs
studied, cigarettes, compared
with alcohol, were more likely
to be used.
27
Other studies (24,34,49) also supported
our findings of relatively high intakes
of vitamin A among pregnant partici­pants.
The 1989 RDA for ascorbic acid
for pregnant women is 70 mg (41).
Overall, our findings regarding the
intakes of the selected B vitamins
(thiamin, riboflavin, niacin, and B1
)
are supported by other studies that
consistently reported intakes of selected
B-complex vitamins as being at least
100 percent ofthe RDAs (12,49) .
Vitamin B6 intakes of our study
participants did not meet the 1989
RDAs for all three groups of women,
a fmding supported by other studjes
(12,49). Women in our study consumed
folate in amounts substantially less than
the RDA. Several studies that reported
average nutrient intakes by pregnant
women, compared with the RDAs,
recorded mean folate intakes below
the RDA (12,24).
Calcium, iron, magnesium, and zinc
intakes for all three groups of women
in our study were less than 83 percent
of the 1989 RDAs. Other studies had
similar fmdings (7, 12,24,49,53). In our
study, phosphorus was the only mineral
that exceeded I 00 percent of the 1989
RDA.
Some of the food composition data­bases
lacked information on nutrients
that may be present in the diets of
pregnant women at levels that are
substantially less than recommended.
These nutrients include vitamins B6
,
B12
, D, and E, and some minerals
(including zinc, magnesium, and
copper) (41). This may explain partially
why the aforementioned nutrients are
among those that are reported to be
consumed consistently in amounts
substantially less than the RDAs.
In our study, body mass index (BMI)
values were considered normal­according
to the guidelines that
consider BMI values between 18.5
and 24.9 as normal (38). BMI is a
28
preferred indicator of nutritional status
because it depends on two commonly
and easily measured aspects of
morphology- weight and height (52).
A large study of3,946 White non­Hispanic
mothers reported BMI values
up to 26 (28) . Similarly, another study
depicted a wider range of BMis from
underweight to obese for their pregnant
participants (43). Pregnancy guidelines
recommend that women of normal pre­pregnancy
weight should gain between
25 and 35 pounds, while underweight
and overweight women should gain
between 35 and 40 and 15 and 25
pounds, respectively (31). The mean
gestational weight gains of the partici­pants
in our study were within the
normal ranges. Other investigators
reported similar mean gestational
weight gains for their participants
(5,26,54,55).
Pre-pregnancy weight-for-height status
is among factors that investigators
have linked with gestational weight
gain (52) . Weights determined at the
first prenatal visit during the first
trimester of pregnancy have been used
to estimate total weight gain and early
gestational weight gain, but these
weights do not necessarily reflect
pregnancy weights. Although average
weight gain in the first trimester is
small relative to that in the second and
thjrd trimesters, individual variation
may be considerable. Total gestational
weight gains may be overestimated by
self-reports or underestimated if based
on weight in the latter part of the first
trimester (52). The Subcommittee on
Nutritional Status and Weight Gain
During Pregnancy suggests that
African-American women should
strive to gain weight at the upper
end of the target weight range (52).
Compared with their counterparts,
women addicted to recreational drugs
are at a higher risk of experiencing a
variety of obstetrical complications that
may increase perinatal morbidity for
mother and child (6). Preventing these
effects should be based on thorough
infom1ation about this segment of the
population-probably via unbiased
longitudinal studies. Prevention of
these deleterious dfects should also be
based on careful medical control of the
nutrition of these mothers, their health
and social conditions during gestation,
and the treatment of their addiction
before and during pregnancy (36).
Acknowledgments
The investigations reported in this
paper were completed as part of the
program project "Nutrition, Other
Factors, and the Outcome of Preg­nancy."
The project was supported by
the National Institute of Child Health
and Human Development of the
National Institutes of Health, through a
grant to the School of Human Ecology
at Howard University in Washington,
DC. The authors thank the following
co-investigators for their valuable
contributions to the research project:
Dr. Ura Jean Oyemade, Dr. 0. Jackson
Cole, Dr. Ouida E. Westney, Dr.
Lennox S. Westney, Ms. Haziel Laryea,
and Dr. Sidney Jones.
Family Economics and Nutrition Review
References
I. Abrams, B. and Newman, Y. 1991. Small-for-gestational-age birth: Maternal
predictors and comparison with risk factors of spontaneous pre-term delivery in the
same cohort. American Journal of Obstetrics and Gynecology 164:785-790.
2. Adams, E.H., Gfroerer, J.C., and Rouse, B.A. 1989. Epidemiology of substance
abuse including alcohol and cigarette smoking. Annals of the New York Academy
of Sciences 562: 14-22.
3. Archie, C.L., Anderson, M.M., and Gruber, E.L. 1997. Positive smoking history
as a preliminary screening device for substance use in pregnant adolescents.
Journal of Pediatrics Adolescence and Gynecology 10(1): 13-17.
4. Bendich, A. 1993. Lifestyles and environmental factors that can adversely affect
maternal nutritional status and pregnancy outcomes. ln C.L. Keen, A. Bendich, and
C.C. Willhite (Eds.) Maternal Nutrition and Pregnancy Outcome. Annals of the
New York Academy of Sciences 678:255-265.
5. Bergmann, M.M., Flagg, E.W., Miracle-McMahill, H.L., and Boeing, H. 1997.
Energy intake and net weight gain in pregnant women according to body mass
index. International Journal of Obstetrics and Related Metabolic Disorders
21 (II): I 0 I 0-10 I 7.
6. Bishai, R. and Koren, G. 1999. Maternal and obstetric effects of prenatal drug
exposure. Clinical Perinatology 26(1):75-86.
7. Brennan, R.E., Kohrs, M.B., Nordstrom, J.W., Sauvage, J.P., and Shank, R.E.
1983. Nutrient intakes of low income pregnant women: Laboratory analysis of
foods consumed. Journal of the American Dietetic Association 83:546-550.
8. Bunout, D. 1999. Nutritional and metabolic effects of alcoholism: Their
relationship with alcoholic liver disease. Nutrition 15(7-8):583-589.
9. Butz, A.M., Lears, M.K., O'Neil, S., and Lukk, P. 1998. Home interventions for
in utero drug-exposed infants. Public Health Nursing 15(5):307-318.
10. Dubowitz, L.M.S., Dubowitz, Y., and Golberg, C. 1970. Clinical assessment of
gestational age in the newborn infant. Journal of Pediatrics 77:1-10.
II. Eisner, Y., Brazie, J.V., Pratt, M.W., and Hexter, A.C. 1979. The risk of low
birthweight. American Journal of Public Health 69:887-893 .
12. Endres, J., Dunning, S., Poon, S.W., Welch, P., and Duncan, H. 1987. Older
pregnant women and adolescents: Nutrition data after enrollment in WI C. Journal
of the American Dietetic Association 87:1011-1016, 1019.
13. Federal Register; Part XII, Friday, August 14, 1987.
14. Fenton, L., Mclaren, M., Wilson, A. , Anderson, D., and Curry, S. 1993.
Prevalence of maternal drug use near time of delivery. Connecticut Medicine
57(10):655-659.
2001 Vol. 13 No. 2 29
30
15. Fisher, S.E., Atkinson, M., Van Thiel, D.H., Rosenblum, E., David, R.,
and Holzman, I. 1981. Selective fetal malnutrition: The effect of ethanol and
acetaldehyde upon in vitro uptake of alpha amino isobutyric acid by human
placenta. Life Science 29:1283-1288.
16. Flynn, A., Miller, S.I., Martier, S.S., Golden, N.L., Sokol, R.J ., and Del
Villano, B.C. 1981 . Zinc status of pregnant alcoholic women: A determinant of
fetal outcome. Lancet I :572-575.
17. Frank, A., Zuckerman, B.S., Amaro, H., Aboagye, K., Bauchner, H. , Fried,
C.L., Hingson, R., Kayne, H., Levenson, S.M., Parker, S., Reece, H., and Vinci, R.
1988. Cocaine use during pregnancy: Prevalence and correlates. Pediatrics
82:888-895.
18. Frederick, J. and Adelstein, P. 1987. Factors associated with low birth weight
of infants delivered at term. British Journal of Obstetrics and Gynecology 85: 1-7.
19. Frederick, J. and Anderson, A.B.M. 1976. Factors associated with spontaneous
pre-term birth. British Journal of Obstetrics and Gynecology 83:342-350.
20. Fried, P.A., Innes, K.S., and Barnes, M.V. 1984. Soft drug use prior to and
during pregnancy: A comparison of samples over a four-year period. Drug and
Alcohol Dependence 13: 161-176.
21. Goodman, L.S. and Gilman, A.G. 1982. Goodman & Gilman's: The
Pharmacological Basis of Therapeutics (7th ed, pp. 560-566). Institute of
Medicine, Washington, DC. National Academy of Sciences.
22. Hans, S.L. 1999. Demographic and psychosocial characteristics of substance­abusing
pregnant women. Clinical Perinatology 26(1):55-74 .
23. Hingson, R., Zuckerman, B., Amaro, H., Frank, D.A., Kayne, H., Sorenson,
J.R., Mitchell, J., Parker, S., Morelock, S., and Tim, P.R. 1986. Maternal
marijuana use and neonatal outcome: Uncertainty posed by self-reports.
American Journal of Public Health 76:667-669.
24. Hunt, l.F. , Murphy, N.J., Cleaver, A.E., Faraji, B., Swendseid, M.E., Coulson,
A.H., Clark, V.A., Laine, N., Davis, C.A., and Smith Jr. , J.C. 1983. Zinc supple­mentation
during pregnancy: Zinc concentration of serum and hair from low
income women of Mexican descent. American Journal of Clinical Nutrition
3 7:572-582.
25. Jacobson, S.W., Jacobson, J.L., Sokol, R.J., Martier, S.S., Ager, J.W., and
Kaplan, M.G. 1991. Maternal recall of alcohol, cocaine, and marijuana use during
pregnancy. Neurotoxicology and Teratology 13(5):535-540.
26. Johnson, A.A., Knight, E.M., Edwards, C.H., Oyemade, U.J., Cole, O.J.,
Westney, O.E., Westney, L.S., Laryea, H., and Jones, S. 1994. Dietary intakes,
anthropometric measurements and pregnancy outcomes. The Journal of Nutrition
l24(6S) :936S-942S.
Family Economics and Nutrition Review
27. Johnson, A.A., Knight, E.M., Edwards, C.H., Oyemade, U.J., Cole, O.J.,
Westney, O.E., Westney, L.S. , Laryea, H., and Jones, S. 1994. Selected lifestyle
practices in urban African-American women: Relationship to pregnancy outcome,
dietary intakes and anthropometric measurements. The Journal of Nutrition
124:963S-972S.
28. Kleinman, J.C. 1990. Maternal weight gain during pregnancy: Determinants
and consequences. NCHS working paper series no. 33. National Center for Health
Statistics, Public Health Service, U.S. Department of Health and Human Services,
Hyattsville, MD, 24 pp.
29. Knight, E.M., Hutchinson, J. , Edwards, C.H., Spurlock, B.G., Oyemade, U.J.,
Johnson, A.A., West, L.W., Cole, O.J., Westney, L.S., Westney, O.E., Manning,
M., Laryea, H., and Jones, S. 1994. Relationships of serum illicit drug concentra­tions
during pregnancy to maternal nutritional status. The Journal of Nutrition
124(6S):973S-980S.
30. Knight, E.M., Johnson, A.A., Spurlock, B.G., West, W.L., and James, H. 1992.
Illicit drug use in pregnancy: Effect of maternal nutritional status and birthweight.
Federation of American Societies for Experimental Biology (Abstract).
31. Kolasa, K.M. and Weismiller, D.G. 1997. Nutrition during pregnancy.
American Family Physician 56:205-212.
32. Kramer, M.S. 1987. Intrauterine growth and gestational duration determinants.
Pediatrics 80:502-511 .
33. Kvigne, Y.L. , Bull, L.B., Welty, T.K., Leonardson, G.R., and Lacina, L.
1998. Relationship of prenatal alcohol use with maternal and prenatal factors in
American Indian women. Social Biology 45(3-4):214-222 .
34. Loris, P., Dewey, K.G., and Poirier-Brode, K. 1985. Weight gain and dietary
intake of pregnant teenagers. Journal of the American Dietetic Association
85:1296-1305.
35. Luke, B., Johnson, T.R.B., and Petrie, R.H. 1993. Maternal-sociodemographic
characteristics. In Clinical Maternal-Fetal Nutrition (pp. 87-120). Little, Brown
and Company, Boston.
36. Martinez-Frias, M.L. 1999. A risk analysis of congenital defects due to
drug intake during pregnancy. Spanish Collaborative Study ofCongenita1
Malformations. Medical Clinics 23;112(2):41-44.
37. Moore, C., Negrusz, A., and Lewis, D. 1998. Determination of drugs of
abuse in meconium. Journal of Chromatography B Biomedical Sciences and
Applications 713(1):137-146.
38. National Heart, Lung and Blood Institute, National Institutes of Health. 1998.
The Evidence Report: Clinical Guidelines on the Identification, Evaluation, and
Treatment of Overweight and Obesity in Adults. U.S. Department of Health and
Human Services, Public Health Service, NIH, Publication No.: 98-4083.
2001 Vol. 13 No.2 31
32
39. National Institute on Drug Abuse (NIDA). NIDA Capsules. 1986 (November).
Rockville, MD.
40. National Institute On Drug Abuse (NIDA). NIDA Capsules. 1989 (June). Drug
Abuse and Pregnancy. DHHS Pub. No. (ADM) 91-1804.
41. National Research Council, National Academy of Sciences, Food and Nutrition
Board. 1989. Recommended Dietary A 1/owances (I Oth ed. ). National Academy
Press, Washington, DC. 284 pp.
42. Nutriplanner 6,000 System. 1987. Practorcare Inc., San Diego, CA.
43. Ogunyemi, D., Hullett, S., Leeper, J., and Risk, A. 1998. Prepregnancy body
mass index, weight gain during pregnancy and perinatal outcome in a rural black
population. Journal of Maternal and Fetal Medicine 7 (4): 190-193.
44. Papoz, L., Eschwege, E., Pequignot, G., Barrat, J., and Schwartz, D. 1982.
Maternal smoking and birth weight in relation to dietary habits. American Journal
of Obstetrics and Gynecology 142:870-876.
45. Pegues, D.A., Engelgau, M.M., and Woernle, C.H. 1994. Prevalence of illicit
drugs detected in the urine of women of childbearing age in Alabama public health
clinics. Public Health Reports I 09(4):530-538.
46. Richardson, G.A., Day, N.L., and McGauhey, P.J. 1993. The impact of
prenatal marijuana and cocaine use on the infant and child. Clinical Obstetrics
and Gynecology 36(2):302-318.
: ,;
47. Robins, L.N. and Mills, J.L. (Eds.). 1993. Effects of in utero exposure to street
drugs. American Journal of Public Health 83: I S-32S.
48. Rubenstein, K.E., Schneider, R.S., and Ullman, E.F. 1972. Homogenous
enzyme immunoassay: A new immunochemical technique. Biochemical and
Biophysical Research Communications 4 7:846-851.
49. Rush, D., Sloan, N.L., Leighton, J., Alvir, J.M., Horowitz, D.O., Seaver, W.B.,
Garbowski , G.C., Johnson, S.S., Kulka, R.A., Holt, M., Devore, J.W., Lynch, J.T.,
Woodside, M.B., and Shanklin, D.S. 1988. The National WIC Evaluation: Evalua­tion
of the Special Supplemental Food Program for Women, Infants, and Children
V. Longitudinal study of pregnant women. American Journal of Clinical Nutrition
48:439-483.
50. SPSS Inc. 1990. SPSS• User's.Guide. SPSS Inc., Chicago.
51 . Stewart, D.E. and Streiner, D. 1994. Alcohol drinking in pregnancy. General
Hospital Psychiatry !6(6):406-412.
Family Economics and Nutrition Review
52. Subcommittee on Nutritional Status and Weight Gain During Pregnancy,
Subcommittee on Dietary Intake and Nutrient Supplements During Pregnancy,
Committee on Nutritional Status During Pregnancy and Lactation; Food and
Nutrition Board, Institute of Medicine, National Academy of Sciences. Substance
Use and Abuse During Pregnancy. 1990. In Nutrition During Pregnancy. Part l.
Weight Gain. Part 2. Nutrient Supplements. National Academy Press, Washington,
DC. pp. 63-95, 96-120, 390-411.
53 . Suitor, C.J.W., Gardner, J., and Willett, W.C. 1989. A comparison offood
frequency and diet recall methods in studies of nutrient intake of low income
pregnant women. Journal of the American Dietetic Association 89: 1786-1794.
54. To, W.W. and Cheung, W. 1998. The relationship between weight in preg­nancy,
birth-weight and postpartum weight retention. Australian New Zealand
Journal of Obstetrics and Gynaecology 38(2): 176-179.
55. Tulman, L., Morin, K.H., and Fawcett, J. 1998. Prepregnant weight and weight
gain during pregnancy: Relationship to functional status, symptoms, and energy.
Journal of Obstetrics, Gynecology, and Neonatal Nursing 27(6):629-634.
56. Vaughn, A.J, Carzoli, R.P. , Sanchez-Ramos, L., Murphy, S., Khan, N., and
Cruu, T. 1993. Community-wide estimation of illicit drug use in delivering women:
Prevalence, demographics, and associated risk factors. Obstetrics and Gynecology
82(1):92-96.
2001 Vol. 13 No.2
33
Sharada Shankar, PhD, MPH
Department of Epidemiology
Johns Hopkins School of Publ ic Health
Ann Klassen, PhD
Department of Health Policy
and Management
Johns Hopkins School of Public Health
34
Influences on Fruit and
Vegetable Procurement and
Consumption Among Urban
African-American Public Housing
Residents, and Potential
Strategies for Intervention
Epidemiological evidence suggests that diets high in fruits and vegetables
provide protective effects from numerous diseases. Data show that consump­tion
of fru its and vegetables is much lower in low socioeconomic groups.
This study assessed the food-purchasing behaviors and barriers to consuming
fruits and vegetables among African-American women living in public housing
in an urban city. Face-to-face data collection methods included interviews of
two focus groups of l 0 women each and structured-questionnaire interviews
of 230 women. The focus groups addressed the issues of barriers to fruit and
vegetable consumption by the families; the structured-questionnaire interviews
focused on food-purchasing and food-preparation behaviors. Results indicated
that the women wanted to increase fruit and vegetable consumption by their
family, but several barriers existed : Cost, poor cooking skills, lack of social
support, and childhood eating patterns. The women mode several key
suggestions for interventions: Stipends for participants, pictures to illustrate
text, older community members to serve as session leaders, and empathetic
and noncondescending teaching styles.
D iets high in fruits and vegetables
have been shown to protect
against an array of diseases,
cancer included (24,25). Carotenoids
and vitamin C protect against cataracts
(26) and oxidation of cholesterol in the
arteries (9). Increased consumption of
fruits and vegetables has been shown to
reduce elevated blood pressure levels
(1), and also to increase significantly
iron absorption, thus minimizing iron
deficiency anemia (10,31).
Both ethnicity and socioeconomic
resources have been linked to variations
in the consumption offruits and
vegetables. Consumption offruits
and vegetables is lower among low­income
populations than among their
counterparts (1 5,27). Additionally,
the intake of fruits and vegetables is
generally lower among African Ameri­cans
than among Whites (1 1, 16,19).
Various factors affect consumption
offruits and vegetables by low-income
families. Intervention approaches must
consider barriers to purchase, prepara­tion,
and consumption as separate yet
interconnected issues. Although
removing barriers to the purchase
and preparation of fruits and vegetables
is a necessary first step, barriers to
consumption must also be addressed.
Family Economics and Nutrition Review
For example, low-income shoppers may
be reluctant to risk scarce dollars on
foods that are unlikely to be consumed
by their families. Moreover, food
patterns of African Americans vary
according to economic, regional, and
social influences of each community.
Mainstays of African-American food
patterns have drawn on eating habits
of several cultures: that of seventeenth
and eighteenth century West Africans,
culture associated with American
Slavery, and the culture of the post­Civil
War rural South (3,4,13).
One focus group identified cost, limited
storage space, time involved in prepar­ing
food, and difficulty in changing
one's own and children's behavior as
major barriers among low-income
White women who lived in housing
projects (21). Some of the barriers to
consuming fruits and vegetables among
low-income women who participated
in the Special Supplemental Nutrition
Program for Women, Infants and
Children (WIC) were unavailability,
time and effort to prepare the foods,
and preferences for other foods (28).
One limitation of existing work in this
area is that data are often collected
from respondents who do not live
within the same community; hence,
shopping experiences could differ.
Also, an overemphasis on data collec­tion
with participants in programs
such as WIC limits our knowledge to
families with very young children.
This study attempts to overcome
these issues by focusing on women
in a wide age range, all living in one
specific community (23) . Therefore,
this explanatory study assessed food­purchasing
behaviors of public housing
residents in one specific area in an
urban city and the barriers they
encountered to consuming fruits
and vegetables.
2001 Vol. 13 No. 2
Methods
Data Collection and Sample
For this exploratory research, we were
interested in both the frequency and
patterns ofbehaviors: such as shopping,
meal planning, and food consumption,
as well as attitudes and beliefs about
foods and dietary practices. The use of
two complementary methods of data
collection, focus group interviews and
more structured questionnaire inter­views,
allows for both qualitative and
quantitative measurement and analyses.
From the questionnaire interview data,
we could determine the prevalence of
certain food behaviors and which
groups within our low-income popula­tion
were most likely to practice these
behaviors. From the more qualitative
focus group discussions, we could gain
insight into the beliefs and attitudes
associated with the reported behaviors.
The use of multiple methods of data
collection, such as those we used,
provides triangulation and strengthens
the external validity of our findings (2).
These findings are crucial in developing
targeted and tailored interventions.
Structured Interviews
We conducted surveys in late 1997 to
assess the food-purchasing behavior
of public housing residents in one area
of an urban city. The food-purchasing
behavior questionnaire consisted of 22
questions and included:
Sociodemographic information
(age, education, employment, and
number of years lived in public
housing).
• Household structure and
composition.
Shopping behaviors including how
often, where (comer stores vs.
supermarket) and who purchased
the food, and whether the food
purchaser made a grocery list before
shopping.
Information on who was responsible
for preparing the food and whether
there was a household main meal
consumed by all the family
members.
Questionnaire items were developed
by the investigators or adapted from a
questionnaire of the Food Marketing
Institute (8) . The Food Marketing
Institute collects data periodically by
telephone interview on food-purchasing
trends, attitudes, and behaviors from
a representative U.S. population. Our
newly developed questionnaire was
pilot-tested among a small number of
respondents.
The face-to-face interviews were
conducted by trained African-American
interviewers who lived in the urban
community. African-American women
ages 18 and older (N=230) who lived in
one of three public housing complexes
were recruited, by "word of mouth,"
to participate. This nonprobability
sampling method, in which initial
participants are used to recruit other
members of a community, is called
"snowball sampling" (2) . A small cash
remuneration was provided to the
participants. The interviews ranged
from 15 to 20 minutes and were
conducted in respondents' homes
or in nearby community centers.
Focus Groups
Two focus group interviews were
conducted, with I 0 women, ages 30 to
65, participating in each session. One
participant was recruited from each
public housing complex within the
targeted political jurisdictions in the
southeastern section of the urban city.
The sessions lasted 2 hours. Each
participant received a remuneration
of food coupons. The focus group
interviews were conducted by a
professional African-American female
consultant. The questions used in the
focus groups were developed using
35
standard focus group methods (18)
to elicit perception of barriers to the
purchase, preparation, and consump­tion
of fruits and vegetables. The
questions were reviewed by several
nutritionists, behavioral scientists,
anthropologists, and health educators.
In addition, the questions were tested
by several target audiences to deter­mine
whether the questions were
pertinent to this community. Themes
used in the focus groups included
preparation, cost, access, information,
and program participation (table 1).
Analysis
From the questionnaires, we calculated
descriptive statistics for the sample's
demographic characteristics, as well
as food-purchasing behaviors. Student
t test and chi-squares were used to
identify differences in food-purchasing
and cooking behaviors by the sample's
demographics. Statistical Analysis
System (SAS) version 6.12 was used
to perform the analysis (22).
The tape-recorded interviews of the
focus groups were later transcribed.
The two authors read the transcribed
material and made independent notes
of themes and patterns. We looked at
clusters of concepts and ideas between
the focus groups (table 1). The theme
that emerged focused on barriers to
fruit and vegetable consumption, as
well as views on behavior-change
programs. Original quotes were
selected as examples, and the
responses that were specific and
based on personal experiences were
given more consideration than vague
and nonspecific responses.
Results
The sample that completed the
structured questionnaire comprised
230 women who were 18 to 91 years
old (table 2). More than half of the
women (56 percent) were less than
36
Table 1 . Focus group themes and questions
Theme 1-Barriers
What are some of the reasons why people do not buy and eat fruits and vegetables?
What are some of the problems in preparing fruits and vegetables?
Do you think cost is an issue for people in your community for eating fruits and
vegetables?
How can we change issues of cost?
Do you think that having access to fruits and vegetables is a problem for people in
your community? How can this problem be resolved?
Do you think that people just have not heard that eating fruits and vegetables are
good for them?
Theme 2-Motivatars
What are the things that motivate people to make a change in their eating habit?
Where do people get information on food? Do they provide information on eating
more fruits and vegetables?
What was the last such information you saw or heard? What made you pay attention
to it?
As a result of it, did you make a change in your behavior in eating more fruits and
vegetables?
Theme 3-Programs
Have you ever participated in a program that was related to improving your health
status?
What specific aspect of this program did you like or did not like?
Do you think your friends and neighbors would participate in a program that
encouraged them to eat more fruits and vegetables?
Where and at what time of day should the program take place?
Who do you think would be a good person to lead the program?
How would you make the program become a part of the community so that it
continued even when the money was gone that started it?
41 years old and had less than a high
school education (55 percent), an~
almost four-fifths (79 percent) were
not working (unemployed, retired, a
student, or a homemaker). Analysis
of the households in which the women
lived showed that most (89 percent)
lived in households of six or fewer
people. The average household
consisted of3.8 people, a somewhat
larger figure than the 1999 national
average of2.5 for African Americans
(29). Most of the women lived in
households with people less than 18
years old (70 percent) and had lived in
public housing for at least 6 years (63
percent). Over one-third of the women
(36 percent) were single parents.
Structured Interviews
Dinner was the main meal for most
of the respondents (72 percent), and
almost all households consume this
meal together (96 percent) (table 3).
Family Economics and Nutrition Review
Use of prepared or "fast" food occurs
at least once a week for 55 percent of
the respondents. One person, usually
the survey respondent, did most of the
shopping (75 percent) and shopped
for food once every other week (31
percent). About two-fifths (41 percent)
of the households plan their meals
before buying food, compared with
cooking whatever is on hand.
Compared with comer or convenience
stores, supermarkets are the main place
for food shopping (94 percent), with
70 percent of respondents shopping at
markets that are within l 0 blocks of
their homes. An equal number of
respondents (50 percent) use and don't
use an automobile to shop. About one­quarter
(22 percent) walk to food
markets some of the time (data not
shown).
Women who eat dinner as a main meal
are significantly older than those whose
main meal is at other times of the day
(44 vs. 38 years old) (table 4). Those
who are living with other adults and
children in their households, and those
who work are both less likely to be the
sole preparer of meals in their home:
34 and 36 percent, respectively.
Patterns of fast-food consumption
vary among these respondents.
Women who live with children in their
households, either as single parents
or with other adults, are significantly
more likely to eat fast food at least once
a week than those without children in
their households. In addition, younger
respondents, and those who currently
work, are also more likely than their
counterparts to eat fast food.
Overall, sociodemographic charac­teristics
of the women did not
significantly affect food-shopping
behavior (table 5). For this sample,
age is the only significant predictor of
shopping frequency, with older women,
2001 Vol. 13 No.2
Table 2. Demographic characteristics of urban African-American women
residing in public housing: Structured interviews
Characteristic
Sample (n)
Women's age (years)
Household size
Years in public housing
Individual characteristics
Age (years)
< 20
21-40
41-60
> 60
Education
Less than 8th grade
8th - 1 1 th grades
High school
Beyond high school
Employment status
Working full- or port-time
Unemployed
Retired/ student/ homemaker
Other/ don't know
Household characteristics
Number of people in household
1-3
4-6
7-10
Number of persons < 18 years in household
None
1-3
4-7
Household composition
Lives alone
Lives with adult(s)
Single parent
Lives with odult(s) and child(ren)
Years in public housing
0-5
6-10
11+
Statistic
230
Mean
43
3.8
13
Percent
6
50
28
16
9
46
35
10
17
34
45
4
47
42
11
30
50
20
15
15
36
34
37
21
42
37
Focus group participants cited
cost as the primary structural
barrier to fruit and vegetable
consumption.
38
Table 3. Cooking and food-purchasing behaviors of urban African-American
women residing in public housing: Structured interviews
Cha racteristic
Sample (n)
Main meal of the day
Dinner
Other
Most people in household eat main meal together
Yes
No
Meal preparer
Self only
Other1
Use of fast-load per week
1-7 times each week
Never/ seldom
Grocery shopper
Self only
Other1
Frequency of food shopping
Once o week or more
Once every 2 weeks
Once o month
As we need food
When most food shopping is done
Beginning of the month
Middle of the month
End of the month
No preference/ anytime
How cooking is planned
Plan before buying
Cook what is on hand
Both
Where most food shopping is done
Supermarket
Yes
No
Corner/ convenience store
Yes
No
Distance to supermarket
Less than 5 blocks
5-10 blocks
More than 1 0 blocks
Cor used to shop
Yes
No
Food received from other sources2
SHARE progrom3
WIC progrom4
Community co-op
Other
None
Statistic
230
Percent
72
28
96
4
79
21
55
45
75
25
26
31
23
20
49
35
4
12
41
52
7
94
6
4
96
37
33
30
50
50
12
24
16
15
45
10ther includes the respondent and another person who shore the responsibil ity.
2A single subject may receive food from more than one category.
3Self-Help and Resource Exchange.
4Women, Infants and Children.
Family Economics and Nutrition Review
Table 4. M~al p~tterns of African-American women1 residing in public housing, by demographic characteristics:
Structured mterv1ews
Main meal Meals made Fast-food used
is dinner by self only once a week or more
Characteri stic Yes No Yes No Yes No
Mean
Age (years) 44 38* 43 41 38 47*
Years in public housing 14 11 13 16 13 15
Percent
Household composition
Lives alone 71 29 97 3 33 67
Lives with adult(s) 76 24 71 29 45 55
Single parent 70 30 89 11 63 37
Lives with adult(s) and child(ren) 71 29 66 34* 60 40*
Employment status
Working 72 28 64 36 71 29
Not working 71 29 82 18* 50 so·
Education
Less than high school 83 17 87 13 33 67
High school or more 70 30 78 22 43 57*
Distance to the supermarket
1-5 blocks 65 35 74 26 56 44
More than 5 blocks 76 24 82 18 54 46
Uses car to shop
Yes 73 27 75 25 58 42
No 70 30 83 17 51 49
1 n=230.
*Women using these meal patterns ore significantly different, based on t Jests (age) and chi-square tests (categorical variables), at p< O.OS.
on average 48 years old, being more
likely to report shopping at least every
week. Frequency of planning before
buying food and using nonpurchased
food (received through WIC or
charitable organizations) are consistent
across the entire sample, with about
half of the respondents reporting these
behaviors.
Focus Groups
Focus group participants cited cost as
the primary structural barrier to fruit
and vegetable consumption. They
identified some fruits and vegetables
2001 Vol. 13 No. 2
as more economical than others but
believed fruits and vegetables overall
were costly, compared with other
foods, especially by volume or portion.
Volume and the ability to provide
family members with a significant
quantity of food were an important
dimension of the cost theme. For
example, grapes and apples were
mentioned often as highly desirable
fruits in terms oftaste but were
impractical, compared with potatoes
prepared as home fries, in terms of
"filling up" the family.
"They [fruits and vegetables}
cost more than some of the other
things we can eat. If you buy
starches, you can stretch them.
Two cucumbers for $1 maybe,
then where is the rest of the
salad? You know you are going
to want more than cucumbers in
your salad. . .. You see, if you
have eight kids, you have to be
able to have enough food for all
of them. Say you buy apples, you
have to buy eight of them or at
least 10. That's quite a big bill
for apples. "
39
Table 5. Food-purchasing behaviors of African-American women1 residing in public housing, by demographic
characteristics: Structured interviews
Plans before Shops at least Uses free
buying food every week food
Characteristic Yes No Yes No Yes No
Mean
Age (years) 42 43 48 41* 43 43
Years in public housing 12 14 13 14 14 13
Percent
Household composition
Lives alone 57 43 41 59 54 46
Lives with adult(s) 29 71 29 71 47 53
Single parent 48 52 20 80 58 42
Lives with adult(s) and child(ren) 47 53 23 77 55 45
Employment status
Working 56 44 15 85 41 59
Not working 46 54 27 73 57 43
Education
Less than high school 39 61 39 61 61 39
High school or more 48 52 24 76 54 46
Distance to the supermarket
1-5 blocks 44 56 30 70 59 41
Less than 5 blocks 49 51 23 77 53 47
Uses car to shop
Yes 49 51 23 77 51 49
No 45 55 28 72 59 41
1 n= 230.
*Women with these food-purchasing behaviors ore significantly different, based on t tests (age) and chi-square tests (categorical variables), at p<O.OS.
40
"I don't buy my fruits or
vegetables unless they are on
sale . . . . You can clip a coupon
for a can good, but you never see
a coupon for fresh fruits and
vegetables."
"We need to think of a way to put
money in the area specifically for
fruits and vegetables. That s all
you can use [those] little green
coupons [referring to food
stamps} for: fruits and
vegetables. You can't buy meat,
you can't buy [anything]. Just
fruits and vegetables every
month. "
Most respondents acknowledged that
their usual meals did not meet their
own standards for nutrition but that it
was often beyond their financial and
emotional skills to plan and prepare
complex meals. Foods such as Oodles
of Noodles~ were mentioned often in
contrast; they were seen as inexp~nsive,
easier to store and prepare rapidly, and
reliably acceptable as a meal to
children.
Low- or no-cost food programs were
discussed as avenues to decrease the
cost of fruits and vegetables but were
seen as a less desirable source of food,
compared with directly purchasing
food. This was in part because of the
uncertain quality and the schedul e and
volume of distribution. It was also
considered less durable because of how
the food was distributed. The method
used tainted the perceived value of the
food. Several respondents described a
program in which local farm trucks
dumped surplus potatoes onto the
ground near the housing complexes.
"They shouldn 't just throw it on
the ground. We are taught not to
eat off the ground. "
Family Economics and Nutrition Review
They [women] asked for
activities to learn and share
menus that would meet several
criteria: Convenience and cost,
health, and children's tastes.
2001 Vol. 13 No.2
"It's like we are animals. It does
something to the way your
children feel. Even though they
know you may. . . get food stamps
but to see you go out there and
get that food [of! the ground}­they
don't understand it. "
Compared with the significance of
cost, only a few other structural barriers
were considered important. Some
respondents, however, did discuss
barriers such as carrying canned fruits
and vegetables home from the store and
freezing or storing sufficient fruits and
vegetables in small apartments.
As women and heads of households,
most participants described themselves
as cooking for others as well as for
themselves; many spoke of the diffi­culty
of balancing the family's and
children's preferences with budgeting
and cooking constraints. They fre­quently
compared their situations to
their parents'; they believed they were
making a conscious decision to allow
their children more choices in foods
than they had been given.
"I think the times we are living in
make a difference. For example,
when I was growing up, if they
put string beans or squash in
front of me, or anything else that
was in season that they could
afford, I ate it . . . . Today's
parents say if they don't like it
'get on up. ' "
"I believe it is an emotional
thing. When I was growing up,
you had to eat what they gave
you. /just thought that was so
mean, and I swore that I wasn 't
going to treat my children like
that. They don't want it, they do
not have to eat it."
"You shouldn 't have to eat fruits
and vegetables if you don 't like
them."
Knowledge of vegetable preparation
techniques was discussed. Many
women believed that there was less
knowledge of cooking techniques in
their communities