Family Economics and Nutrition Review [Volume 12, Number 2]

              CENTER FOR NUTRITION POLICY AND PROMOTION
Research Articles
3 Dietary Changes in Older Americans From 1977 to 1996:
Implications for Dietary Quality
Shirley A. Gerrior
15 Identifying the Poor and Their Consumption Patterns
Deanna L. Sharpe and Mohamed Abdel-Ghany
Commentary
26 Cruciferous Vegetables and Human Breast Cancer: An Important
Interdisciplinary Hypothesis in the Field of Diet and Cancer
Ron Jevning, Mark Biedebach, and Raj en Anand
Research Briefs
31 Diets of Individuals Based on Energy Intakes From Added Sugars
Shanthy A. Bowman
39 Household Expenditures on Vitamins and Minerals by Income Level
Mark Lino, Julia M Dinkins, and Lisa Bente
44 The Diet Quality of American Indians: Evidence From the Continuing Survey of
Food Intakes by Individuals
P. Peter Basi otis, Mark Lino, and Raj en Anand
Research Summaries
47 Supplemental Security Income Program Participation by Noncitizens
51 A Dietary Assessment of the U.S. Food Supply
55 Who Gained the Most During the 1990's Expansion?
57 The Food-at-Home Budget: Changes Between 1980 and 1992
Regular Items
62 Federal Statistics: Homeownership ..
64 Journal Abstracts
66 USDA Food Plans: Cost of Food at Home
67 Consumer Prices
UNITED STATES DEPARTMENT OF AGRICULTURE
Volume 12, Number 2
1999
itor
ia M. Dinkins
atures Editor
rk Uno
naging Editor
ne W. Fleming
ntributors
an C. Courtless
ce W. Klein
ily Economics and Nutrition Review is
tten and published each quarter by the
nter for Nutrition Policy and Promotion,
. . Department of Agriculture, Washington, DC.
Secretary of Agriculture has determined
t publication of this periodical is necessary
he transaction of the public business
uired by law of the Department.
is publication is not copyrighted. Contents
y be reprinted without permission, but credit
Family Economics and Nutrition Review
uld be appreciated. Use of commercial or
e names does not imply approval or consti­endorsement
by USDA. Family Economics
Nutrition Review is indexed in the following
abases: AGRICOLA, Ageline, Economic
· erature Index, ERIC, Family Studies,
IS, and Sociological Abstracts.
mily Economics and Nutrition Review is
sale by the Superintendent of Documents.
bscription price is $12.00 per year ($15.00
foreign addresses). Send subscription
ers and change of address to Superin­dent
of Documents, P.O. Box 371954,
sburgh, PA 15250-7954. (See subscription
m on p. 71 .)
iginal manuscripts are accepted for publi­tion.
(See "guidelines for authors• on back
ide cover.) Suggestions or comments con­rning
this publication should be addressed
Julia M. Dinkins, Editor, Family Economics
d Nutrition Review, Center for Nutrition
licy and Promotion, USDA, 1120 20th St.
, Suite 200 North Lobby, Washington, DC
36. Phone (202) 606-4876.
e Family Economics and Nutrition Review
now available on (http://www.usda.gov/cnpp)
e Internet (see p. 69).
Center for Nutrition Policy and Promotion
Research Articles
3
15
Dietary Changes in Older Americans From 1977 to 1996:
Implications for Dietary Quality
Shirley A. Gerrior
Identifying the Poor and Their Consumption Patterns
Deanna L. Sharpe and Mohamed Abdel-Ghany
Commentary
26 Cruciferous Vegetables and Human Breast Cancer:
An Important Interdisciplinary Hypothesis in the Field
of Diet and Cancer
Ron Jevning, Mark Biedebach, and Rajen Anand
Research Briefs
31
39
44
Diets of Individuals Based on Energy Intakes From Added Sugars
Shanthy A. Bowman
Household Expenditures on Vitamins and Minerals by Income Level
Mark Lino, Julia M. Dinkins, and Lisa Bente
The Diet Quality of American Indians: Evidence From the
Continuing Survey of Food Intakes by Individuals
P. Peter Basiotis, Mark Lino, and Rajen Anand
Research Summaries
47 Supplemental Security Income Program Participation by Noncitizens
51 A Dietary Assessment of the U.S. Food Supply
55 Who Gained the Most During the 1990's Expansion?
57 The Food-at-Home Budget: Changes Between 1980 and 1992
Regular Items
62 Federal Statistics: Homeownership
64 Journal Abstracts
66 USDA Food Plans: Cost of Food at Home
67 Consumer Prices
Volume 12, Number2
1999
2
Would you like to publish in Family Economics and Nutrition Review?
Family Economics and Nutrition Review will consider for publication articles concerning economic and
nutritional issues related to the health and well-being of families. We are especially interested in studies
about U.S. population groups at risk-from either an economic or nutritional perspective. Research
may be based on primary or secondary data as long as it is national or regional in scope or of national
policy interest. Articles may use descriptive or econometric techniques.
Family Economics and Nutrition Review has a new feature: Research Briefs. We define Research Briefs
as short research articles. Our guidelines are found on the back inside cover of each issue.
We invite submission of Research Briefs; manuscripts may contain findings previously presented at
poster sessions if not published in proceedings (except for abstract).
Manuscripts may be mailed to Julia M. Dinkins, Editor, Family Economics and Nutrition Review,
Center for Nutrition Policy and Promotion. See guidelines on back inside cover for complete address.
Family Economics and Nutrition Review
Shirley A. Gerrior
Center for Nutrition Policy
and Promotion
1999Vol.l2No.2
Research Articles
Dietary Changes in Older
Americans From 1977 to 1996:
Implications for Dietary Quality
Older people are a rapidly growing segment of the U.S. population. In 1996
persons over age 65 made up 12 percent of the population. As a result, more
attention is being paid to their nutritional well-being and health, particularly
regarding the possible link between nutrition and the develop11ent of chronic
disease. The U.S. Department of Agriculture's (USDA) 1977-78 Nationwide
Food Consumption Survey and the 1989-91 and 1994-96 Continuing Survey
of Food Intakes by Individuals were used to examine the dietary quality of
Americans over age 65 and to evaluate changes in their food and nutrient
intakes from 1977-78 to 1994-96. Results showed that the largest changes
were decreased consumption of whole milk, followed by red meat and eggs,
and increased consumption of sweetened beverages, followed by grain mixtures
and snacks, and reduced-fat milks. In general, the same nutrients were below
the Recommended Dietary Allowances (ADA) during both periods. In 1994-
96, intakes of vitamin E and zinc were below the ADA for men and women.
Future increases in the consumption of whole grains, nonfat or reduced-fat
milks, and vegetables and decreases in sweetened beverages and fats will
help improve overall diet quality and help reduce the risk of chronic diseases
associated with poor eating patterns.
ore attention is being paid
to the nutritional status and
nutrition-related health
needs of older Americans,
as wen as to the relationship between
nutrition and the development of older
Americans' many chronic diseases.
Older people are a rapidly growing
segment of the U.S. population. In
1996, 12 percent of the U.S. population
was over age 65; in 1900, 4 percent of
the population was over age 65 ( 18, 19 ).
By 2050 the older population win more
than double, with most of the growth
occurring between 2010 and 2030 when
members of the baby-boom generation
enter their elderly years ( 18 ). Among
older Americans, women outnumber
men by 6 miiiion (18.9 versus 12.9
minion) because of a higher death rate
among older men ( 18,19 ). It is therefore
important to examine the dietary quality
of Americans over age 65 and to evaluate
changes in their food and nutrient intakes
over time. A better understanding of
the dietary quality and food and nutrient
intakes of elderly Americans win help
identify those at nutritional risk and those
whose nutritional status may be improved
by preventive nutritional interventions.
Accordingly, policymakers and nutrition
professionals will need to target food
and nutrition programs for elders, estab­lish
policies related to food fortification,
and develop nutrition- and health-related
strategies that help elders better meet
the nutritional challenges associated
with aging.
3
The importance of proper nutrition
throughout the life cycle is key in deter­mining
quality of life. Proper nutrition
helps to diminish health problems and
physiological decline associated with
poor diets and poor health habits over
a lifetime. And in the later years, good
nutrition helps to maintain a more
healthful lifestyle and one with greater
independence. An increasing percentage
of elderly people, as they age, face
chronic, limiting illnesses or conditions
such as arthritis, poor dentition, reduced
gastrointestinal functioning, diabetes,
osteoporosis, senile dementia, and
depression. These conditions may result
in an overall decrease in their intake
of food energy and essential nutrients.
These conditions will, as well, impair
an older person's ability to purchase and
prepare nutritious foods, the result of
which may be dependency on others for
help performing daily activities ( 6,11 ).
In general, data on food intake from
national dietary surveys (2,14), as well
as others (4,12), suggest that older
Americans consumed less food than
required to meet recommendations for
food energy and nutrients. Many older
Americans, including the presumably
healthy, have reported nutrient intakes
below the recommendation for food
energy, vitamin E, vitamin B6, calcium,
magnesium, and zinc ( 14 ). These low
intakes, however, may be a problem of
the survey methods used, that is, under­reporting
of the foods consumed.
Recent findings from the Third National
Health and Nutrition Examination Survey
(NHANES III) indicate that 18 percent
of the men and 28 percent of the women
underreported their energy intakes (2).
Underreporting of energy intake was
highest in women and persons who were
older, overweight, or trying to lose weight.
4
Also, intakes of vitamins, minerals, fiber,
and macronutrients were significantly
lower and, in general, paralleled energy
intakes. Although underreporting of food
quantities and food energy has been
observed among the elderly, it is not a
unique problem of this segment of the
population.
Despite underreporting by the elderly,
they may actually eat less as they age
because of a number of factors, including
a decline in physical activity and a
decrease in appetite. Also, surveys show
that energy intakes are consistently low
for the elderly (2,3,12,14,15), suggesting
a real decline in food and nutrient intakes
with age. This contrasts with a higher
mean energy intake by the general popu­lation
that is seen in the more recent
surveys where additional probes were
used for purposes of enhancing recall
(2,15). Thus despite the limitation of
survey respondents underreporting food
quantities and food energy, dietary survey
data are useful when assessing the dietary
quality and food and nutrient intakes of
the elderly, and the data provide impor­tant
information on the nutritional status
of the elderly.
This study examines the dietary quality
of Americans over age 65 and evaluates
changes in their food and nutrient intakes
from 1977-78 to 1994-96. Nutrient
intakes and consumption of major food
groups and subgroups from 1977-96
are examined in terms of current dietary
guidance. By increasing our knowledge
of the dietary behaviors of older people
and observing changes in these behaviors
over time, we can more effectively evaluate
nutrition education efforts and determine
future directions for nutrition intervention,
the goal of which is to improve the quality
of life for this segment of the population.
Methods
Data Source
Data for this study were from the USDA's
Nationwide Food Consumption Survey
of1977-78 (NFCS 77-78) (16,17) and
the 1989-91 and 1994-96 Continuing
Survey of Food Intakes by Individuals
(CSFII) ( 14, 15 ). The NFCS 77-78
included individuals selected from
stratified-area probability samples of
noninstitutionalized households in the
48 conterminous States. For the NFCS
77-78 four waves were conducted, one
for each season, and each on a different
sample of participants ( 8,9 ). Individual
dietary data for 3 consecutive days were
obtained through a mix of a 24-hour
recall and 2-day food records.
The CSFII for 1989-91 and 1994-96
comprise a nationally representative
sample of noninstitutionalized persons
residing in the United States for each
year of the 3-year data sets. For the 1989-
91 CSFII, the USDA used a 1-day recall
and a 2-day dietary record to collect
food intake data for 3 consecutive days
(14). The 1989-91 CSFII included an
all-income and a low-income sample,
which were combined through sample
weights. For the 1994-96 CSFII, USDA
collected 2 nonconsecutive days of
dietary data for individuals of all ages.
The data were collected between January
1994 and January 1997; in-person inter­views
were used to collect the 24-hour
recalls. Only the first day of dietary
intake data was used because day-1
data (for each of the surveys used)
were collected using the 24-hour recall
method. Methods of data collection
used on subsequent days were not as
comparable. Research has indicated that
food intake data based on 1-day dietary
intakes provide reliable intakes by groups
of people ( 1 ). Thus, to best examine
Family Economics and Nutrition Review
changes over time from surveys with
different numbers of days of dietary
information, this study compared esti­mates
of food and nutrient intake among
the surveys based on only the frrst day's
data collected for each individual.
The data were collected from selected
individuals in each household. The method
for collecting the 24-hour recall was
modified from previous surveys to
improve the collection of dietary intake
data and included more questions that
probed the respondents' recollection ( 15 ).
Sample
In this study, older men and older women
made up 10 to 12 percent of the U.S.
population between 1977 and 1996: men,
4 to 5 percent; women, 6 to 7 percent.
The sample selected for analysis con­sisted
of persons older than age 65 who
provided a valid 1-day, 24-hour recall
of dietary intake. For each of the three
surveys, the USDA developed sample
weights to adjust for survey response
and for other vagaries of sample selection.
Use of weighted data provides results
that are more characteristic and gener­alizable
to the U.S. population.
Nutrient Analysis
This study examined food energy,
15 nutrients, and dietary components.
Nutrient data were not available in the
NFCS 77-78 for saturated fat, choles­terol,
folate, vitamin E, zinc, dietary
fiber, and sodium. The nutritive value
of the foods the elderly said they con­sumed
was calculated using the USDA's
Nutrient Data Bank and survey data­bases
for 1977-78, 1989-91, and 1994-
96. Average nutrient intakes for l day
were computed for these three periods.
Nutrient intakes as a percentage of the
1989 Recommended Energy Allowance
(REA) or Recommended Dietary Allow­ances
(RDA) were used in this study.
1999 Vol. 12 No.2
They were derived by dividing an indi­vidual's
intake by the REA or RDA for
the appropriate age/gender group.
Food Analysis
Ten major food groups (used by USDA)
and 27 food items that reflect the total
diet were analyzed (table 1). For the
CSFII surveys, USDA has developed a
Food Grouping System for separating
mixtures into their component parts ( 14 ).
However, in this study, for purposes of
comparability between the NFCS and
CSFII, food mixtures were not separated
into individual ingredients. For example,
grain and meat mixtures were placed
into a grain or meat mixture category
based on the primary ingredient (e.g.,
a macaroni and cheese mixture was
assigned to the grain mixture group; the
macaroni was not assigned to the grain
group, and the cheese was not assigned
to the milk group). Average intake in
grams for each of the food groups and
subgroups was calculated from 1-day
recall for 1977-78 and 1994-96.
Statistical Analysis
Descriptive statistics were derived using
the Statistical Package for the Social
Sciences (SPSS) ( 13 ). Tests for signifi­cance
were not performed. The differ­ences
in the sampling methods of the
surveys and the use of sample weights
limit the degree to which the survey
data can be compared.
For this article, a "trend" was defined
as a "change" in the consumption of a
food or in nutrient intake. For a given
food group (or food), a trend existed
when the mean intakes of the food
group or food rose or fell continually
from 1977-78 through 1989-91 and to
1994-96. Further analysis with more
complex methods (i.e., time trends or
time series analysis) may reveal
additional information.
Both older men and
women increased their
intakes of vitamins A,
C, and Bs; calcium;
iron; phosphorus;
and magnesium.
5
Table 1. Percentage change in average intake (grams per day), 1977-78 to 1994-96, for Americans over age 65
NFCS 1977-78 NFCS 1977-78 CSFII 1994-96 CSFII 1994-96
Men Women Men Women Men Women
Sample size 1,514 2,167 1,101 1,026
----------------------------------Grams -----.--------------------------- ----------Percent change -------
Total meat 223 155 204 151 -9 -3
Red meat 82 51 39 24 -52 -53
Luncheon meats 21 11 21 13 0 18
Poultry 27 24 22 22 -19 -8
Fish 10 9 15 12 50 33
Mixtures 67 46 102 77 52 67
Total milk and milk products 253 216 260 211 3 -2
Total fluid milk 210 157 185 148 -12 -6
Whole milk 109 75 43 29 -60 -61
Reduced-fat milks 53 51 93 70 65 37
Cheese 14 17 15 13 8 -24
Milk desserts 24 20 37 28 54 40
Eggs 39 21 21 15 -46 -29
Legumes, nuts, and seeds 28 16 42 26 50 63
Total grains 232 182 301 232 30 27
Breads and rolls 65 49 61 48 6 -2
Other baked goods 65 45 48 35 -26 -22
Cereals and pasta 67 55 102 71 52 29
Grain snacks 3 3 8 6 167 100
Mixtures 31 31 64 57 106 84
Total vegetables 248 219 252 210 2 -4
White potatoes 72 57 66 47 -8 -18
Tomatoes 29 29 35 30 21 3
Dark-green vegetables 11 12 16 18 45 50
Deep-yellow vegetables 12 15 15 11 25 -27
Other vegetables 123 106 120 104 -2 -2
Total fruits 169 177 214 195 27 10
Citrus 66 74 79 78 20 6
Other fruits 103 103 130 113 26 10
Fats and oils 15 12 17 14 13 17
Table fats 9 6 7 5 -22 -17
Salad dressing 6 4 9 9 50 125
Sugars and sweets 30 20 23 18 -23 -10
Nonalcoholic beverages 617 571 629 16 10
Coffee 443 374 344 -5 -8
Tea 117 141 147 12 4
Carbonated soft drinks 41 41 93 195 127
Fruit drinks 16 17 39 165 129
Note: Food item totals may not equal food group totals because of rounding.
6
Family Economics and Nutrition Review
Table 2. Mean nutrient intakes by gender for older Americans over age 65, 1977 to 1996
19771 19771 1989-912 1989-912 1994-963 1994-963
Men, Women
Sample size 1,037 1,726
Mean
Food energy (kcal) 1,910 1,401 1,823
Total fat (gm) 88.3 61.6 68.2
Saturated fat (gm) 23.6
Cholesterol (gm) 284
Dietary fiber (gm) 17.5
Vitamin A (IU) 6,33'8 6,015 ~.505
Vitamin C (mg) 87 87 J10
Vitamin B6 (mg) 1.63 1.30 1.9
Vitamin E (mg) 8.7
Folate (!lg) '309
Calcium (mg) 709 555 733
Iron (mg) 12.7 9.4 16.3
Phosphorus (mg) 1,194 897 1,204
Magnesium (mg) 257 202 287
Zinc (mg) 13.0
Sodium (mg) 3,275
~Mean intakes per individual in a day, 1-day data, 1977-78 NFCS.
3
Mean intakes per individual in a day, !-day data, 1989-91 CSFII.
Mean intakes per individual in a day, 1-day data, 1994-96 CSFII.
Results
Changes in Average Daily Nutrient
Intakes, 1977-96
From 1977-96, older men's average
intakes of food energy decreased (1 ,910
to 1,854 kcal); older women's intakes
remained essentially unchanged ( 1 ,401
to 1,407 kcal) (table 2). These intakes
are below the 1989 REA for men (2,300
kcal) and women (1,900 kcal) (7). Both
older men and women increased their
intakes of vitamins A, C, and B6; calcium;
iron; phosphorus; and magnesium. They
decreased their intake of total fat: men
by 20 grams and women by 11 grams.
From 1989-96 intakes of dietary fiber
1999 Vol. 12 No.2
increased slightly; intakes of folate,
saturated fat, and cholesterol decreased.
Also, intakes of zinc and sodium for
men were lower in 1994-96 than they
were in 1989-91.
Average Intakes as a Percentage
of Recommendation, 1977-96
Older Americans' diets failed to meet
the 1989 REA for food energy for each
of the survey years, with women's intake
less than 75 percent of the REA (table 3).
Both older men and women exceeded
the recommendation for total fat (107 to
129 percent) and for saturated fat (103
to 118 percent) for all years. However,
total fat and saturated fat intakes as a
Women Men Women
1,377 1,101 1,026
1,392 1,854 1,407
51.9 68.3 50.2
17.6 22.5 15.9
194 256 185
13.5 18.6 14.0
7,651 8,613 6,464
102 107 95
1.5 1.98 1.53
7.1 8.9 6.7
240 298 222
596 778 587
12.0 16.6 12.6
927 1,214 940
224 291 229
8.6 11.0 8.3
2,263 3,179 2,344
percentage of recommendation declined,
an indication that in the past decade
some progress was made in achieving
the goals for fat intake. Whereas both
older men and women met the choles­terol
recommendation (300 milligrams
or less per day), only older women met
the sodium recommendation (2,400
milligrams per day). Older men and
women failed to meet the dietary fiber
recommendation of 25 grams per day:
intakes ranged from 54 to 74 percent
of the recommendation. Older men and
women also failed to meet 100 percent
of the RDA for vitamin B6, vitamin E,
calcium, magnesium, and zinc. In 1994-
96, calcium and zinc intakes for the
7
Table 3. Average intake as percentage of recommendation by gender for older Americans over age 65, 1977
to 1996
19771 1989-912 1994-963
Women Men Women Women
Sample size 1,726 780 1,377 1,026
Percent
Foodenergl 83 74 79 73 74
Total fat 129 129 112 112 107
Saturated fat 118 113 103
Cholesterol 95 65 62
Dietary fiber 70 54 74 56
Vitamin A 127 150 170 191 181 183
Vitamin C 144 146 182 169 172 160
Vitamin B6 82 81 96 93 99 95
VitaminE 87 89 88 82
Folate 154 133 143 128
Calcium 89 69 92 74 96 75
Iron 127 94 163 120 125
Phosphorus 149 112 116 117
Magnesium 73 72 80 82
Zinc 72 70
Sodium 94 98
1Mean intakes per individual in a day, !-day data, 1977-78NFCS.
ZM:ean intakes per individual in a day, 1-day data, 1989-91 CSFll.
3Mean intakes per individual in a day, 1-day data, 1994-96 CSFll.
4Nutrient recommendations are based on the 1989 Recommended Dietary Allowances; total fat iss 30 percent of total calories; saturated fat is < 10 percent of
total calories. Dietary fiber is based on daily intake of 25 grams; sodium, 2,400 milligrams; and cholesterol, S 300 milligrams.
elderly women were 75 percent or less
of the RDA. Despite these shortfalls,
intakes of calcium, vitamin B6, and
magnesium were higher in 1994-96 than
they were in 1977-78 and contributed to
meeting a greater percentage of the rec­ommendation.
The Percentage of Older Americans
With Diets Meeting 100 Percent
of the Recommendation, 1977-96
The percentage of older men and older
women with intakes of food energy that
met 100 percent ofthe REA was low:
8
25 and 17 percent, respectively in 1994-
96, and it essentially remained the same
over the 20-year period (table 4). From
1977-78 to 1994-96, the percentage of
older men and women meeting 100 per­cent
of the recommendation for intakes
of total fat, vitamin B6, and iron increased
notably. In 1989-91 and 1994-96, a
higher percentage of older women than
older men met 100 percent of the recom­mendation
for nutrients that need to
be consumed in moderation: Total fat,
saturated fat, cholesterol, and sodium.
The percentage of older men and
women meeting 100 percent of the
recommendation for mineral intake
(calcium, magnesium, and zinc) was
low throughout the study period.
Changes in Average Intake
(in grams per day) from 1977-96
Total meat products. In 1994-96 older
Americans ate less total meat and 50
percent ate less red meat (beef and pork)
than they did in 1977-78 (table 1). Not
expected was the decrease during this
period in poultry consumption by older
Americans (19 percent less for men and
Family Economics and Nutrition Review
Table 4. Percentage of older Americans by gender over age 65, with diets meeting 100 percent of the
recommendation for intake, 1977 to 1996
19771 19771 1989-912 1989-912 1994-963 1994-963
Men Women Men Women Men Women
Sample size 1,037 1,726 780 1,377 1,101 1,026
Percent
Food energy 4 26 16 21 13 25 17
Total fat 13 18 36 39 37 43
Saturated fat 40 43 43 50
Cholesterol 62 83 67 80
Dietary fiber 20 6 20 11
Vitamin A 40 46 49 51 50 51
Vitamin C 57 58 64 67 62 61
Vitamin B6 26 27 37 36 42 38
Vitamin E 23 26 27 28
Folate 58 60 57
Calcium 17 21 40 22
Iron 37 51 76 56
Phosphorus 55 58 80 62
Magnesium 19 23 28 27
Zinc 19 17
Sodium 61 55
1
Mean intakes per individual in a day, !-day data, 1977-78 NFCS.
~ean intakes perindividualin a day, !-day data, 1989-91 CSFII.
4
Mean intakes per individual in a day, !-day data, 1994-96 CSFII.
Nutrient recommendations are based on the 1989 Recommended Dietary Allowances; total fat is 5. 30 percent of total calories; saturated fat is < 10 percent of
total calories. Dietary fiber is based on daily intake of 25 grams; sodium, 2,400 milligrams; and cholesterol, 5. 300 milligrams.
8 percent less for women). The average
intake offish and meat mixtures, how­ever,
increased substantially. Because
meat mixtures may include appreciable
amounts of red meat or poultry, actual
consumption of these discrete foods
may be higher than the individual food
items indicate.
Total milk products. A noticeable shift
from whole milk to reduced-fat milks
occurred between 1977-78 and 1994-
96, with the elderly drinking 60 percent
1999 Vol. 12 No. 2
less whole milk and 37 to 65 percent
more reduced-fat milks. Despite this
shift in milk types during this period,
both older men and women consumed
less fluid milk overall.
Eggs; legumes, nuts, and seeds. From
1977-78 to 1994-96, egg consumption
decreased for both elderly men and
women-more so for the men than for
the women: -46 versus -29 percent. This
is in contrast to the increased consump­tion
of legumes, nuts, and seeds: 50 per­cent
for men and 63 percent for women.
Total grains. Older men and women ate
more grain products, especially grain
mixtures and snacks (i.e., pizzas and
pretzels), in 1994-96 compared with
1977-78. They also ate more cereals and
pastas, with the change in men's intake
double that of women's: 52 versus 29
percent.
Total vegetables and total fruits. Total
vegetable intake between 1977 and
1996, on average, remained relatively
constant for elderly Americans-they
ate less white potatoes but more tomatoes
9
Older Americans may
be at risk for micro­nutrient
deficiencies ....
10
and deep-green vegetables. Older men
and women consumed about 50 percent
more dark-green vegetables and older
men about one-fourth more deep-yellow
vegetables and one-fifth more tomatoes.
Also, older men and women ate more
total fruit, with their intake of both citrus
and noncitrus fruits higher in 1994-96
than in 1977-78.
Fats and oils. Elderly Americans ate
slight! y more fats and oils in 1994-96
than they did in 1977-78, with a shift
from table fat to salad dressings. For
both men and women, their use of table
fats (margarine and butter) in 1994-96
was about one-fourth less than their use
in 1977-78.
Nonalcoholic beverages. While older
Americans ate less sugar and sweets in
1994-96 than they did in 1977-78, their
consumption of carbonated soft drinks
and fruit drinks increased appreciably,
counterbalancing the positive effects of
consuming less sugar and sweets.
Discussion and Conclusions
Older Americans appear to be moving
towards dietary guidance and closer to
the 1995 Dietary Guidelines for Americans
by incorporating nutrition education
messages into healthful eating behaviors.
From 1977-78 to 1994-%, older Americans
made considerable changes in their diets.
In 1994-96, their consumption of red
meat, eggs, and sugars and sweets was
lower than it was in 1977-78. Their
consumption of legumes, total grains,
and fruits was higher in 1994-96 than it
was in 1977-78. This selection of food
provided less fat, saturated fat, choles­terol,
zinc, and sodium to their diet and
more vitamins A and C, folate, dietary
fiber, calcium, and other bone-related
nutrients.
Despite these dietary changes, average
intakes of food energy, dietary fiber,
vitamins B6 and E, calcium, magnesium,
and zinc were lower than recommenda­tions.
In particular, low calcium intakes
are a concern for both older men and
women, especially in terms of bone
health. The declining use of fluid milk
products, coupled with the increasing
use of soft drinks and fruit drinks is a
troubling trend. The consumption of
soft drinks and fruit drinks is likely to
displace more nutritious foods (e.g.,
milk products and fruits) from the diet
and negatively affect diet quality.
Also, low intakes of dietary fiber and
zinc require attention. While the intake
of dietary fiber may be due to the low
food energy intakes of this sample, these
intakes are considerably below intakes
expected of individuals consuming the
recommended servings of fruits, vegeta­bles,
and whole-grain foods, based on
the Food Guide Pyramid.
Older Americans have included more
of these foods in their diets over the past
10 years. They, however, must continue
to make more appropriate food choices
and work harder to meet nutrient recom­mendations
by increasing the number of
servings of fruits, vegetables, and whole
grains consumed, as well as by increasing
their servings of milk and meat products.
For example, including plenty of fortified
cereal foods in the daily diet may counter­balance
the loss of zinc from red meat
and may also make important contribu­tions
to their intakes of magnesium and
folate-thereby improving dietary quality.
Overall, the low intake of food energy
may prevent the older American from
achieving the balance of foods needed
for optimal diet quality, as indicated
by the many nutrients below the
recommendation.
Family Economics and Nutrition Review
In addition, the older Americans' marginal
and low dietary intakes of many minerals
and vitamins are a concern. Older Ameri­cans
may be at risk for micronutrient
deficiencies not only from low dietary
intakes but also from other non-food
factors, such as the ability to buy and
prepare food, the presence of disease,
or limited income. While the marginal
and low dietary intakes of some nutrients
(vitamin E, calcium, magnesium, and
zinc) in this study are suggestive of
clinical deficiencies, such a risk has
not been confirmed by biochemical or
clinical markers. However, studies using
biochemical markers provide some
evidence regarding the link between low
dietary intake and biochemical status.
The Boston Nutritional Status Survey
of the Elderly and related work ( 10,12)
suggest that older people, even in a
relatively well-off and generally well­nourished
population, may not be getting
as much vitamins as they need. For
example, plasma levels of pyridoxal
phosphate and other measures of vitamin
B6 status have been shown to decline
with age. Erythrocyte activity (ETK-AC),
a biochemical marker of thiamin, has
shown a significant relationship between
thiamin intake and blood levels. Using
this marker, the researchers in the Boston
study ( 12) categorized 5 percent of the
study population as deficient, but the
study noted that a correlation is more
likely to exist between ETK-AC value
and supplemental thiamin than between
ETK-AC and dietary thiamin. Also,
intake of riboflavin has been shown to
have a significant effect on the erythrocyte
glutathione reductase activity1 coeffi­cient
(EGR-AC) in the population regard­less
of gender-with a deficiency noted
in 5 to 16 percent of elderly people in tech­nologically
advanced countries ( 10 ).
1 A biochemical marker of riboflavin activity.
1999 Vol. 12 No.2
For folate, the concentration of folate in
erythrocytes is considered a better indi­cator
of folate stores in the tissue. Serum
levels accurately reflect recent dietary
intake. Currently, the level of homo­cysteine
is linked to a person's folate
status, with serum homocysteine levels
correlated closely with folate intakes
less than 400 Jlg per day.
As with vitamins, the dietary intake of
minerals also has shown a correlation
with biochemical markers. Phosphorus
intakes relate closely to blood phosphorus
levels as does dietary iron and its storage
to plasma ferritin levels levels ( 12). The
requirements for calcium in terms of
bone mineral loss over time have been
linked to the biochemical marker, 25-
hydroxy vitamin D-the levels of which
are lower in older persons than in
younger persons ( 10).
An older person's risk for inadequate
dietary intake is well established. The
results presented from this study empha­size
the fact that the quality of older
Americans' diet continues to need
improvement. Nutrition intervention
strategies need to be developed that
improve nutrient intake for the older
American. These strategies should
emphasize the total diet and overall diet
quality; they should help reduce the risk
of chronic diseases associated with poor
eating patterns. A diet needs to be low
in fat and saturated fat and contain foods
that provide adequate amounts of essen­tial
minerals, vitamins, and dietary fiber.
For older people, efforts should be
targeted to increase their intakes of
food energy, dietary fiber, vitamin E,
folate, calcium, magnesium, and zinc.
Limitations of Study
This study has two major limitations in
terms of the implications presented: (1)
the survey data and (2) the use of the
RDA versus the Dietary Reference
Intake (DRI) for assessment of dietary
quality.
Survey data. The survey design and
nutrient databases, underreporting by
survey respondents, and the use of
24-hour recall data are included in this
limitation. The use of different surveys
and nutrient databases may make the
data of the earlier years less comparable
to the data of later years, especially in
terms of the intake of fat and cholesterol
and possibly folate. The nutrient data
for the later surveys reflect improved
data as well as changes in the nutrient
content of foods that are attributable to
new varieties and species, to new fortifi­cation
levels, and to changes in the
practices of the food industry.
Dietary intake was assessed using data
from 24-hour recalls. Such data are poor
indicators of a given person's usual diet
but are more useful to characterize a
group's intake when the sample size is
sufficient (5). When providing dietary
information, survey respondents tend
to underreport consumption of certain
foods, especially those foods high in
fat and calories; they also tend to over­report
consumption of foods (e.g., fruits)
that are high in nutrients. Underreporting
has decreased somewhat in more recent
surveys (CSFII 94-96 and NHANES
III) because more probes and collection
methods have been added. Underreporting,
however, remains a problem in certain
subgroups, primarily women and persons
who are older, overweight, or on a diet
to lose weight. Additional research is
11
needed to determine the extent of
underreporting of foods consumed, food­preparation
methods and ingredients,
food quantities, and the effect of under­reporting
on estimates of food and nutrient
intakes (2).
RDA vs. DRI. Adopted by the Food
and Nutrition Board, Dietary Reference
Intakes (DRI) represent the new approach
to providing quantitative estimates of
nutrient intakes for use in a variety of
settings, thus replacing and expanding
on the past 50 years of periodic updates
and revisions of the RDA. The new DRI
differ in amounts and age categories
from the 1989 RDA and include three
new categories of reference values:
Adequate Intake (AI), the Estimated
Average Requirement (EAR), and the
Tolerable Upper Intake Level (UL).2
The design of this study does not allow
calculation of the percentage of AI for
calcium or calculation of the percentage
of RDA for phosphorus, magnesium,
folate, or vitamin B6. However, older
Americans' mean intake of these nutrients
as a percentage of their DRI differs
from their mean intake as a percentage
of the 1989 RDA. Compared with the
higher calcium AI value (1,200 mg/d
for men and women age 51 and older),
mean intake for both men and women
met a much lower percentage of the
DRI than for the 1989 RDA. This is
also observed for the mean intakes of
magnesium and folate, with a higher
DRI magnesium RDA value ( 420 mg/d
for men and 320 mg/d for women age
51 and older) and a higher DRI folate
RDA value (400 l!g/d for men and
2-rhe EARs and A Is for the elderly may reflect a
greater variability in requirement, especially for
nutrient-related energy expenditures (20 ).
12
women age 51 and older), respectively
than for the 1989 RDA. The mean intake
of phosphorus met a higher percentage
of the DRI (700 mg/d for men and
women age 51 and older) than of the
1989 RDA. Also, mean intakes of
vitamin B6 met a higher percentage of
the DRI (1.7 mg/d [RDA] for men and
1.5 mg/d [RDA] for women) than of the
1989 RDA, with older men and women
in 1994-96 exceeding the DRI.
Family Economics and Nutrition Review
References
1. Basiotis, P.P., Welsh, S.O., Cronin, F.J., Kelsay, J.L., and Mertz, W. 1987. Number
of days of food intake records required to estimate individual and group nutrient
intakes with defined confidence. The Journal of Nutrition 117(9 ): 1638-1641.
2. Briefel, R.R., Sempos, C.T., McDowell, M.A., Chien, S., and Alaimo, K. 1997.
Dietary methods research in the third National Health and Nutrition Examination
Survey: Underreporting of energy intake. American Journal of Clinical Nutrition
65(suppl): 1203S-1209S.
3. Brown, J.E., Tharp, T.M., Dahlberg-Luby,E.M., Snowdon, D.A, Ostwald, S.K.,
Buzzard, I.M., Rysavy, D.M., and Wieser, M.A. 1990. Videotape dietary assessment:
Validity, reliability and comparison of results with 24-hour dietary recalls from
elderly women in a retirement home. Journal of the American Dietetic Association
90:1675-1679.
4. Cid-Ruzafa, J., Caulfield, L.E., Barron, Y., and West, S.K. 1999. Nutrient intakes
and adequacy among an older population on the eastern shore of Maryland: The
Salisbury eye evaluation. Journal of the American Dietetic Association 99:564-571.
5. Levine, E. and Guthrie, J. 1997. Nutrient intakes and eating patterns of teenagers.
Family Economics and Nutrition Review 10(3):20-35.
6. Mahan, L.K. and Escott-Stump, S. (Eds.). 1996. Krause's Food, Nutrition and
Diet Therapy (9th ed.). W.B. Saunders Company. Philadelphia.
7. National Research Council, Subcommittee on the Tenth Edition of the RDAs,
Food and Nutrition Board. 1989. Recommended Dietary Allowances (10th ed.).
National Academy Press, Washington, DC.
8. Peterkin, B., Rizek, R.L., and Tippett, K.S. 1988 (January/February). Nationwide
Food Consumption Survey. Nutrition Today, pp. 18-24.
9. Rizek, R.L. 1978. The 1977-78 Nationwide Food Consumption Survey. Family
Economics Review 4:3.
10. Russell, R.M. 1997. New views on the RDAs for older adults. Journal of the
American Dietetic Association 97:515-518.
1999 Vol. 12 No. 2 13
11. Ryan, AS., Craig, L.D., and Finn, S.C. 1992. Nutrient intakes and dietary patterns
of older Americans: A national study. Journal of Gerontology 47(5):M145-M150.
12. Sahyoun, N. 1992. Nutrient intake by the NSS elderly population. In S.C. Hartz,
I.H. Rosenberg, and R.M. Russell (Eds.), Nutrition in the Elderly, The Boston
Nutritional Status Survey (pp. 31-44 ). Smith-Gordon and Company Ltd, London.
13. SPSS, Inc. 1996. SPSS Base 7.5 for Windows. SPSS, Inc., Chicago, IL.
14. U.S. Department of Agriculture, Agricultural Research Service. 1995. Food and
Nutrient Intakes by Individuals in the United States, 1 day, 1989-91. Tippett, K.S.,
Mickle, S.J., Goldman, J.D., Sykes, K.E., Cook, D.A., Sebastian, R.S., Wilson, J.W.,
and Smith, J. NFS Report No. 91-2.
15. U.S. Department of Agriculture, Agricultural Research Service. 1998. CD-ROM
Documentation: 1994-96 Continuing Survey of Food Intakes by Individuals and
1994-96 Diet and Health Knowledge Survey. Riverdale, MD.
16. U.S. Department of Agriculture, Human Nutrition Information Service. 1983.
Food Intakes: Individuals in 48 States, Year 1977-78. Nationwide Food Consumption
Survey 1977-78. Report No.l-1. Hyattsville, Maryland.
17. U.S. Department of Agriculture, Human Nutrition Information Service. 1984.
Nutrient Intakes: Individuals in 48 States, Year 1977-78. Nationwide Food Consumption
Survey 1977-78. Report No.l-2. Hyattsville, Maryland.
18. U.S. Department of Commerce, Bureau of the Census, Economics and Statistics
Administration. 1995. 65 Plus in the United States. Statistical Brief. [on-line]
www.census.gov/socdemo/www/agebrief.html.
19. U.S. Department of Commerce, Bureau of the Census. 1997. Statistical Abstract
of the United States, 1997. [117th ed.].
20. Yates, A.L., Schlicker, S.A., and Suitor, C.W. 1998. Dietary Reference Intakes:
The new basis for recommendations for calcium and related nutrients, B vitamins,
and choline. Journal of the American Dietetic Association 98:699-706.
14 Family Economics and Nutrition Review
Deanna L. Sharpe
University of Missouri-Columbia
Mohamed Abdei-Ghany
University of Alabama
1999 Vol. 12 No.2
Identifying the Poor and
Their Consumption Patterns
We used three increasingly restrictive measures to differentiate the poor
from the nonpoor. Findings show that little difference exists among those
classified as poor by any of the three poverty measures used. However,
compared with the nonpoor, the poor--by all three measures-were
younger, had more children under age 18, and had fewer vehicles and
wage earners. The poor were more likely than the nonpoor to be Black,
be single, have a high school education; and to rent or reside in govemment
housing. By using the most restrictive measure of poverty, we found signifi­cant
spending differences between the poor and nonpoor for food at home,
housing, health, transportation, and other expenses.
ITJ he current poverty measure
in the United States consists
of a set of thresholds that
are compared with house­hold
before-tax annual income and
adjusted for household composition.
Originally developed in the 1950's, the
thresholds are based on the cost of a
minimal diet times a multiplier of three
to account for other expenses (4). The
thresholds are adjusted periodically for
inflation. In measure and design, how­ever,
the thresholds have not changed
substantively since their inception.
Recently the official poverty measure
has faced two broad criticisms. One:
given the growth in in-kind and income
transfer programs since the War on
Poverty began in the 1960's, before-tax
income no longer reflects accurately a
household's economic resources (9).
Two: the multiplier used in the official
poverty measure is based on the as­sumption
that households still spend
one-third of their budget on food. Over
time, however, the percentage of the
household budget spent on food has
declined, while the percentage of the
household budget spent on other items
(i.e., housing, health care, transporta­tion,
and child care) has increased (2).
Alternative measures of poverty have
been proposed to address these criticisms.
For example, adding the value of in-kind
transfers to before-tax income can help
make income a better estimate of house­hold
economic resources ( 5 ). However,
inaccurate reporting of income can cause
practical problems when researchers
use social survey data. Expenditure­based
poverty measures have been
proposed as another alternative to the
current income-based measure
(9,14,15).
The Consumer Expenditure Survey
(CE), which provides extensive infor­mation
on the expenditures of American
consumers, is a logical data set to use
to develop an expenditure-based measure
of poverty. Lino ( 10) has proposed
using total expenditures reported in
the CE along with household income
to assess poverty status. The measure
based on total expenditures that is
available in the CE, however, has been
15
criticized as being biased because the
purchase price of high-cost durables
is included in the measure when the
purchase is made. A measure that
focuses on regular out-of-pocket
expenses, so-called total outlays, has
been proposed as an alternative to
total expenditures ( 13 ).
The purpose of this paper is twofold.
First, it examines the characteristics of
households classified as poor by three
increasingly restrictive measures of
poverty: The official income-based
measure, a total-expenditure measure
plus the current income-based measure,
and a total-outlay measure plus the
previous two measures. Second, this
paper examines the spending behavior
of households classified as poor by the
most restrictive of these three measures.
Comparisons between the poor and the
non poor are made.
Method
Data and Sample
Data are from the Interview portion of
the 1994 CE, which is conducted by the
Bureau of the Census for the Bureau
of Labor Statistics (BLS). An ongoing
survey, the CE gathers information
on expenditures, income, and major
sociodemographic characteristics of
consumer units1 in the civilian non­institutionalized
population. BLS uses
a rotating panel design to survey about
5,000 consumer units each quarter.
1 A consumer unit is defined as either all merri>ers
of a household who are related by blood, marriage,
adoption, or other legal arrangement; a finan­cially
independent person living alone or as a
roomer; or two or more persons living together
and making joint expenditure decisions. In this
paper, the terms consumer unit and household
are used interchangeably.
16
Consumer units contribute five con­secutive
quarters of data; about 20 per­cent
of the sample is replaced by new
participants each quarter ( 16 ).
The CE treats each interview as an
independent observation ( 16). We used
CE weights to adjust the sample to
reflect the population. We also omitted
from this analysis consumer units
whose household head was not White
or Black, incomplete income reporters?
and consumer units with negative levels
of household before-tax income, total
expenditures, or total outlays.3 The
resulting unweighted sample size
was 16,367.
Poverty Measures
We used three measures of poverty in
this study. The first measure compares
household before-tax annual income to
the official poverty guidelines. Adjust­ments
for household size and the age
2Complete income reporters have provided infor­mation
on major income sources such as wages
and salaries, self-employment income, and Social
Security income for the consumer unit for the
previous 12 months. However, in addition to these
major income sources, annual before-tax income
also includes amounts received during the pre­vious
12 months by members of the consumer
unit for Supplemental Security Income, unem­ployment
compensation, workers' compensation,
veterans' payments, public assistance, interest
and dividend income, pension income, rental
income, alimony and child support received,
and value of food stamps. A consumer unit that
reports a value of zero for all sources of income
is classified as an incomplete income reporter.
lntese omissions were based largely on pragmatic
reasons. Of the total sample, less than 4 percent
were neither White nor Black. Incomplete income
reporters and consumer units reporting negative
levels of income or spending may have insufficient
or incorrect income or expenditure data, thus
limiting our ability to classify them appropriately
as poor or non poor. Our decision to exclude
incomplete income reporters and those with
negative income or expenditures reduced the
sample by about 2 percent.
of the household head in one- or two­person
families are reflected in these
guidelines. We classified those house­holds
reporting income below the rele­vant
threshold as poor. This measure
we termed the "single-hurdle" measure
because only this one hurdle or standard
must be cleared for the household to be
considered poor.
Variations in income receipt and the
inability or unwillingness of survey
respondents to report completely and
accurately how much income was
received can cause income to be an
unreliable measure of household eco­nomic
resources (7, 12,13 ). Further, in
the CE, a consumer unit is classified
as a complete income reporter when
values have been reported for major
income sources-even though informa­tion
may not have been provided for all
income sources (7,16). No attempt is
made in the CE to impute income when
it is missing. Obviously, when consumer
income is understated (whether by error
or intent), the consumer unit can be
classified as poor when it is not.
Given these problems in measuring
income, researchers have used total
expenditures as a proxy for income
( 1,12). The theoretical basis for this
substitution is the permanent income
hypothesis. It suggests that consumers
try to maintain a given level of con­sumption
over time and are relatively
unresponsive to transitory increases
and decreases in income. Thus: com­pared
with measures of annual income,
annual total expenditures are a better
representation of consumption patterns
over the lifespan (7).
One drawback of using total expendi­tures
instead of income when assessing
poverty status is that a household might
appear to be poor on the basis of total
Family Economics and Nutrition Review
expenditures when it is simply saving
rather than spending ( 11 ). To overcome
this drawback, Lino ( 10) suggested
using total expenditures in addition to,
rather than instead of, income when
assessing whether a household is above
or below the poverty thresholds. The
second measure of poverty we used in
this study classified a household as
poor if its before-tax annual income
and its total annual expenditures were
below the relevant dollar values of the
official poverty thresholds. This measure
we termed a "double hurdle" because
two criteria must be met for a house­hold
to be considered poor.
In the CE, the purchase price of con­sumer
durables is included in total
expenditures when the purchase is made.
Purchase of high-cost durables (i.e.,
vehicles) can bias the total-expenditure
measure upward. Conversely, the CE
excludes principal payments on home
mortgages from total expenditures.
(Home mortgage interest is included
in total expenditures.) Thus the total­expenditure
measure can be biased
downward for homeowners who make
mortgage payments. Rogers and Gray ( 13)
have proposed an alternative measure
called "total outlays" that is designed
to capture the "regular out-of-pocket
outlays of consumers." This measure
adds principal payments on home mort­gages
and on financed vehicles to total
expenditures and subtracts the purchase
price of financed vehicles. To construct
our third measure of poverty, we com­puted
total outlays for the sample. Then
we classified a household as poor if its
before-tax annual income, total expendi­tures,
and total outlays were below the
relevant official poverty guidelines.
This measure we termed "triple hurdle"
because three criteria must be met for
the household to be classified as poor.
1999 Vol. 12 No.2
Introduction of each additional hurdle
makes the definition of poverty more
restrictive. Consequently, of those
classified as poor by the single-hurdle
measure, not all remain classified as
poor when the double-hurdle measure
is imposed. Of those designated as poor
using the double hurdle, not all remain
classified as poor when the triple-hurdle
measure is applied.
Variables
Variables used only in the descriptive
statistics included household head over
age 64, household size, number of vehicles,
being a renter, government housing,
before-tax income, total outlays,
Supplemental Security Income (SSI),
welfare benefits, and food stamp value
(table 1). Variables used only in the
regression analysis included region of
residence, Interview quarter, and the
poverty measure. Remaining variables
were used in both the descriptive statistics
and the regression analysis.
Sociodemographic variables used as
independent variables in the regression
analysis included age, education, and
race of the household head (defined
as the husband in husband-wife house­holds);
number of children less than
age 18; household type; number of
earners; and region of residence. These
variables were selected to control for
differences in need and preferences.
Reference categories for the categorical
variables were having a high school
education, being White, being a husband­wife
household, and residing in a rural
area. The CE does not report region for
rural residents in order to preserve the
privacy of survey respondents. Thus
we used rural residence as the reference
category for the four urban regions, a
common practice when using CE data.
... when the most
restrictive definition
of the poor is used ...
the poor and non poor
have significantly
different spending
patterns for food at
home, housing, health
care, transportation,
and other expenses ....
17
Economic variables used in the regres­sion
analysis included total expendi­tures
and the poverty measure: a
categorical variable coded 1 if the
household was classified as poor by
the most restrictive measure of poverty
(the triple hurdle), 0 otherwise. Total
expenditures were used as a proxy for
income to address the problems of
income measurement in the CE ( 1 ).
The quarter4 in which the interview
took place was included in the study to
control for possible seasonal effects in
spending behavior. Quarter 4 was the
reference category.
We used eight expenditure categories
as dependent variables in the regression
analyses: Food at home, food away
from home, apparel and apparel services,
housing, transportation, adjusted trans­portation,
5 health, and other. The "other"
category included expenditures on
tobacco, alcohol, education, reading,
entertainment, personal care, personal
4Quarter I included months I through 3; quarter 2,
months 4 through 6; quarter 3, months 7 through
9; and quarter 4, months I 0 through 12.
5 Analysis using the summary variable for trans­portation
in the CE indicated that the poor spent
more for transportation, all else equal. Because
this result may have been related to the way the
CE treats transportation expenditures, an alterna­tive
measure of transportation expenses was
constructed. This alternative measure was con­ceptually
similar to the total-outlay measure
suggested by Rogers and Gray ( 13 ). Expenditures
for public transportation were excluded from
the summary measure of transportation, while
principal payments for financed new and used
cars and trucks were added, and the purchase
price of financed new and used cars and trucks
was subtracted. Specifically, adjusted transporta­tion
was the sum of annualized expenditures for
net outlay for new and used cars and trucks; other
vehicles; gas and motor oil; vehicle finance charges,
maintenance and repairs; vehicle insurance;
vehicle rental, leases, licenses, and other charges;
principal payments for financed new and used
cars and trucks; less the purchase price of
financed new and used cars and trucks.
18
Table 1. List of variables
Variable
Sociodemo~raphic
Age of household head
Household head over age 64
Education of household head
Race of household head
Number of children under age 18
Number of earners
Number of vehicles
Household type
Household size
Region of residence
Renter
Government housing
Economic
Before-tax income
Total expenditures
Total outlays
Supplemental Security Income
Welfare benefits
Food stamp value
Measurement/description
Continuous
Categorical
1 if true; 0 otherwise
Categorical
Less than high school
High school (reference category)
Some college
College
Categorical
Black
White (reference category)
Continuous
Continuous
Continuous
Categorical
Husband-wife (reference category)
Male single parent
Female single parent
One person
Other
Continuous
Categorical
Northeast urban
Midwest urban
South urban
West urban
Rural (reference category)
Categorical
1 if rent; 0 otherwise
Categorical
1 if have; 0 otherwise
Continuous
Continuous
Continuous
Continuous
Continuous
Continuous
Family Economics and Nutrition Review
Table 1. List of variables (Coot' d)
Variable
Qtherindependent
Poverty measure
Interview quarter
Expenditure
Food at home
Food away
Apparel and apparel services
Housing
Health
Transportation
Adjusted transportation
Other
Total expenditures
1999 Vol. 12 No.2
Measurement/description
Categorical
1 if poor by triple-hurdle measure;
0 otherwise
Categorical
Quarter 1
Quarter2
Quarter 3
Quarter 4 (reference category)
Food and beverages purchased and
prepared by the consumer unit for its
own use
Food and beverages purchased by the
consumer unit at restaurants, cafes,
and fast-food establishments
Expenditures for shoes, clothing, sewing
supplies, laundry, and dry cleaning
Expenditures for mortgage interest,
property tax, maintenance, repairs,
insurance and other related expenses,
rent, utilities, household operations,
and home furnishings
Expenditures for health insurance,
medical services, prescription drugs,
and medical supplies
Expenditures for new and used cars and
trucks, gasoline, maintenance and repairs,
vehicle insurance, and vehicle rental
Expenditures for transportation plus
principal payments for financed new
and used cars and trucks Jess purchase
price of financed cars and trucks
Expenditures for tobacco, alcohol,
education, reading, entertainment,
personal care, personal insurance, cash
contributions, and miscellaneous goods
and services
Sum of expenditures for food at home,
food away, apparel and apparel services,
housing, health, transportation, and
other goods and services
insurance, cash contributions, and
miscellaneous items. Multiplying the
total dollar amount spent on each of
the eight expenditure categories by four
annualized the quarterly expenditure
data.
Statistical Analysis
To compare the characteristics of the
poor and nonpoor, we computed
weighted means for relevant variables
for four groups: Those classified as
poor when the single-hurdle measure
of poverty was used, those classified as
poor when the double-hurdle measure
of poverty was used, those classified as
poor when the triple-hurdle measure of
poverty was used, and those not classi­fied
as poor by any of the three measures.
To compare spending patterns of the
poor and nonpoor, we included in each
regression analysis a dummy variable
indicating the household was poor by
the most restrictive measure of poverty.
Because all expenditure categories
used in this study had a relatively low
percentage of zero spending, ordinary
least squares (OLS) regression results
were not biased ( 8 ). Including the
poverty measure in the regressions
indicated whether significant differ­ences
existed in the spending behavior
of the poor and non poor after control­ling
for the age, education, and race
of the household head, the number of
children less than age 18, household
type, number of earners, region of
residence, and quarter in which the
interview was conducted.
19
20
Nearly haH of the sample
that is classified as
poor by the income
threshold measure is
no longer classified
as poor when ... total
expenditures and total
outlays must also be
below the poverty
thresholds.
Findings
Comparison of Characteristics
When the single-hurdle measure of
poverty was used, 15 percent of the
sample was classified as poor.6 When
the second hurdle was imposed, about
half as many-7.4 percent of the sample­was
still classified as poor. Imposing
the third hurdle reduced the percentage
of poor to 7.2 percent.
We found little difference among the
characteristics of those classified as
poor by any of the three measures
(table 2). This result is not surprising:
the double- and triple-hurdle measures
identify a subset of those initially
identified as poor by the single-hurdle
measure. In general, the poor households
were headed by someone who was about
45 years old. Average household size
is close to 3. Compared with households
classified as poor by the single-hurdle
measure, households classified as poor
by either the double- or triple-hurdle
measure had a slightly larger household
size with more children less than age 18
but with fewer vehicles and earners.
While households classified as poor by
all three poverty measures reported less
income than expenditures, the difference
between before-tax income and total
expenditures or total outlays is greatest
for those classified as poor by the single­hurdle
measure. However, this group
reported the smallest average dollar
amount of transfer income (SSI, welfare,
food stamp) among the poor, suggesting
credit or unreported income sources
make up the difference. Nearly one­fourth
of the poor household heads
6-rhis percentage compares favorably with the
14.5 pe~~treported for the U.S. population in
the StatiStical Abstract of the United States 1996
Table No. 736, "Persons Below Poverty Level '
by Race and Family Status 1979 to 1994," p. 47S.
were Black; over 40 percent had less
than a high school education and lived
alone. Over half of the poor were renters;
about 10 percent lived in government
housing.?
Those classified as not poor by all
three poverty measures were slightly
older and more likely to be living in
husband-wife households than were
those classified as poor. Relatively
few had children under 18 years of
age, suggesting these households were
preparing their older children for adult­hood.
This group, on average, had the
largest number of earners and vehicles.
Mean before-tax household income was
$40,424 with mean total expenditures
and mean total outlays of $32,804 and
$32,629, respectively. Interestingly, a
few in this group reported receipt of
welfare benefits and housing support.
Perhaps some of the household units in
this group include one or more members
(i.e., an elderly parent living with an
adult child or a parent and child living
with the child's grandparents) who
could qualify for government transfers.
Comparison of Spending Behavior
Expenditure categories used in this
study focused on the basic necessities
of food, clothing, shelter, transportation,
and health care. Remaining expenditure
categories were classified as other.
Results of the OLS regressions indicate
that when the most restrictive definition
of the poor is used (the triple-hurdle
measure), the poor and nonpoor have
7These results differ somewhat from Lino's ( 10).
The ?ifferences are likely the result of focusing
on different groups for analysis. Lino studied
households with children. Our study includes
households with and without children. Conse­quently,
in our study, the average age of the
household head is older, and the household size
is smaller.
Family Economics and Nutrition Review
Table 2. Means of selected variables for the poor and nonpoor
Poverty measure for the poor
Single Double Triple
hurdle hurdle hurdle
(15% of (7.4% of (7.2% of
sample) sample) sample) Non poor
Means
Age of household head 45.54 44.50 44.55 48.56
Household size 2.51 2.71 2.69 2.51
Number of children <18 years 0.96 1.19 1.19 0.64
Number of vehicles 0.98 0.66 0.63 2.15
Number of earners 0.77 0.65 0.63 1.41
Before-tax income $6,913.45 $7,183.25 $7,163.55 $40,423.93
Total expenditures $14,124.59 $8,259.19 $8,165.59 $32,803.81
Total outlays $14,185.75 $8,419.23 $8,274.01 $32,628.84
Supplemental Security Income $360.07 $482.33 $474.81 $96.12
Welfare benefits $793.66 $1,144.77 $1,162.43 $63.28
Food stamp value $693.46 $1,033.65 $1,043.91 $42.80
Percent
Household head >64 years 24 22 22 22
Household head Black 23 28 28 8
Household head education
<High school 41 50 50 17
High school 27 27 27 31
Some college 23 20 19 25
College 8 3 3 27
Household type
Husband-wife 26 21 20 57
Male single parent I 1 1 1
Female single parent 17 22 22 4
One person 41 41 42 27
Other 15 15 15 11
Renter 55 60 61 30
Government housing 7 10 11 1
1999 Vol. 12 No.2
significantly different spending pat­terns
for food at home, housing, health
care, transportation, and other expenses
(table 3). No significant differences
in spending between the poor and non­poor
were found for food away from
home, apparel and apparel services,
and adjusted transportation.
With two exceptions-transportation
and other expenses-the poor spent
less than the nonpoor spent. Using the
summary measure of transportation in
the CE, we found that the poor spent,
on average, $1,904 more than did the
nonpoor for transportation. Additional
investigation suggested this unexpected
result was due to differential expendi­tures
for new and used cars and trucks
and for public transportation. When
transportation expenses were adjusted
to remove expenses for public transpor­tation
and the net outlay for fmanced
vehicles (comparable to the adjustment
made to total expenses to compute total
outlays), the spending difference between
poor and non poor ceased to be statisti­cally
significant.8 Findings also indi­cated
that the poor spent, on average,
$752 more than the nonpoor spent for
other expenses (tobacco, alcohol, edu­cation,
reading, entertainment, personal
care, personal insurance, cash contribu­tions,
and miscellaneous).
8When amount spent for public transportation
is the dependent variable in an ordinary least
squares regression that has the same set of inde­pendent
variables as are used in this article, the
poor spent almost $117 more per year than the
nonpoor. The t value for this result is 2.366,
si~ficant at the 0.5 level. However, the adjusted
R for this model is quite low at 8 percent.
21
Table 3. Regression analysis of spending pattern differences between poor and non poor
Variable
Total expenditure
Age of household head
Education of household head
<High school
Some college
College
Household head Black
Number of children <18 years
Household type
Male single parent
Female single parent
One person
Other
Number of earners
Region
Northeast urban
Midwest urban
South urban
West urban
Interview quarter
Quarter 1
Quarter2
Quarter3
Poverty measure 1
Constant
Adjusted R2
I Triple-hurdle measure.
* p<.OI.
** p<.OOI.
***p<.OOOI .
Discussion
The income threshold measure of poverty
is an absolute standard designed to
reflect ability to meet basic needs. A
household is poor if its before-tax
income is below the threshold. As a
measure of poverty, it is simple to
22
Expenditure category
Food at home Food away
Betas
0.02*** 0.03***
13.43*** 3.11 **
133.01 * -48.71
55.34 68.42
136.86** 264.93***
-38.90 -147.11 *
539.85*** 116.49***
-348.96 322.25
-429.04*** -110.70
-1130.89*** -349.88***
-228.84*** -64.75
219.68*** 119.12***
499.11*** 286.50***
-86.19 99.44
44.28 126.65*
307.83*** 267.98***
31.57 -64.38
-2.34 -43.37
27.57 -95.53
-295.91 *** -89.39
1495.88*** 605.43***
.32 .23
implement and easy to understand.
It provides an objective measure for
assessing qualification for welfare
benefits. But there are problems with
its use. To the extent that the income
people report is incomplete or incorrect,
a household may be classified errone­ously
as poor. Before-tax income may
Apparel and
apparel services Housing
0.04*** 0.18***
-0.94 1.00
-91.10 -405.17*
219.05*** 448.61 **
400.42*** 2402.77***
207.36*** 52.24
68.24*** 496.13***
-530.10* -1542.99*
79.89 -188.09
-188.73*** -1 033.44***
-47.09 -563.65***
58.00* -96.58
256.93** 2728.70***
173.87* 1154.17***
142.99* 1324.54***
35.95 2587.91 ***
424.86*** -203.76
-157.42** -131.60
-114.52* -91.95
-67.01 -1588.21***
-60.09 1895.52***
.24 .50
not reflect accurately a household's
economic resources if the household
receives transfer payments. Calculation
of the thresholds has also been criticized.
At present, the cost of families' basic
needs is calculated as three times the
cost of a minimal diet, adjusted for
household composition. However, in
Family Economics and Nutrition Review
Table 3. Regression analysis of spending pattern differences between poor and nonpoor (Cont'd)
Expenditure category
Variable Health Transportation
Betas
Total expenditure 0.03*** 0.44***
Age of household head 35.98*** -44.39***
Education of household head
<High school -62.27 802.17
Some college -1.97 -1257.50***
College -76.65 -4791.83***
Household head Black -483.82*** 759.84*
Number of children <18 years 56.77 -739.05***
Household type
Male single parent -720.62* 1940.50
Female single parent -252.75 1071.07*
One person -636.75*** 2555.43***
Other -309.16*** 1525.53***
Number of earners -250.61 *** -778.24***
Region
Northeast urban -479.95*** -3132.72***
Midwest urban -290.16*** -1298.98***
South urban -145.14 -1515.81***
West urban -392.34*** -2559.26***
Interview quarter
Quarter 1 30.97 -351.27
Quarter2 193.33* 132.89
Quarter3 31.06 415.11
Poverty measure1 -615.67*** 1904.35***
Constant -82.76 -1685.66**
Adjusted R2 0.14 0.53
1Triple-hurdle measure.
* p<.Ol.
** p<.OOI.
*** p<.OOOI.
the years since the threshold was imple­mented,
the percentage of food in the
budget has declined, making the multi­plier
too small, and other expenditure
categories (i.e., housing, health care,
transportation, and child care) now vie
with food for consideration as "basic
expenses" (2).
An expenditure-based measure of
poverty proposes several advantages
over an income-based measure. It allows
a wide definition of basic expenses to
be considered. Consumers are often
more willing to disclose expenditures
than income. Expenditures tend to be
free of the transitory increases and
1999 Vol. 12 No.2
Adjusted Other
transportation expenses
0.22*** 0.26***
-12.93* -8.19
238.30 -327.92
-306.83 468.05*
-1522.25*** 1664.28***
-95.36 -349.61
-335.41 *** -538.42***
1206.68 879.92
64.39 -170.38
451.39* 784.26***
310.70 -312.04
72.71 728.63***
-1889.52*** -138.58
-607.02* 247.85
-705.68** 22.49
-1355.15*** -248.07
-94.56 132.01
79.40 8.51
153.65 -171.74
97.84 751.84**
255.40 -2168.32***
0.31 .52
decreases that can occur with income.
But consumers can choose to spend less
than income and thus be misclassified
as poor when an expenditure-based
measure is used. Including the net
purchase price of high-cost durables
in expenditures when the purchase is
made, as is done in the CE, can bias
23
results. Further, while income repre­sents
a measure of resources that can
be used to secure items needed for
survival, expenditures simply reflect
past purchasing decisions. Nothing is
known about either the quantity or
quality of items purchased. Families and
individuals are designated as poor by
using any expenditure-based measure
without reference to any objective stand­ard
of need (which the official income
thresholds attempt to reflect by using
cost of a minimal diet as a basis).
Another practical concern is that, in
its present form, the CE does not have
a sufficient sample size to provide
detailed regional analysis of poverty
(3).
In this study, as in Lino's study (10),
the use of a poverty measure based on
annual before-tax income and annual
total expenditures mitigates the limita­tions
present when either income or
expenditures are used alone. This study
carries this approach one step further
by imposing yet another criterion for
comparison with the poverty thresholds­total
outlays. Use of total outlays
provides limited correction for the
problem of having the purchase price
of a high-cost durable included in total
expenditures. The resulting poverty
measure is restrictive. Nearly half of
the sample that is classified as poor
by the income threshold measure is
no longer classified as poor when the
additional criterion is imposed-that
total expenditures and total outlays must
also be below the poverty thresholds.
24
Summary and Implications
The purpose of this paper was to com­pare
the characteristics and spending
behavior of households classified as
poor and non poor by using three in­creasingly
restrictive measures: An
income-based measure (the current
official poverty measure), the income­based
measure plus a total expenditure­based
measure, and the income- and
total expenditure-based measure plus a
total outlay-based measure. Findings
indicate that little difference exists
among those classified as poor by any
measure. There are several differences
in the characteristics of those classified
as poor by any measure and those not
classified as poor by any measure.
After controlling for several socio­demographic
variables, we found that
spending patterns for food at home,
housing, health, transportation, and
other expenses were significantly
different for those classified as poor
by the triple-hurdle measure, the most
restrictive measure of poverty, and the
non poor. With the exception of trans­portation
and other expenses, the poor
spent less than the non poor spent.
It is beyond the scope of this research
to propose which poverty measure
should be used. Selection of a poverty
measure must account for many factors,
including the purposes for which the
measure will be used, national living
standards, and social norms regarding
the ways in which, and the degree to
which, those deemed poor should be
helped. However, we found relatively
small differences in the characteristics
of those classified as poor by either
the double-hurdle or the triple-hurdle
measure. This result suggests that
while correcting for the cost of high­priced
durables can be defended on
logical and theoretical grounds, differ­entiating
between total expenditures
and total outlays may make little
practical difference.
Comparing both income and expendi­ture
levels with the official poverty
thresholds offsets the limitations
present when using a measure of income
or expenditures alone to identify the
poor. This approach helps minimize
the possibility of misclassifying as
poor those who underreport income
but have high expenditures or those
who have high incomes but choose a
relatively low level of spending. The
resulting poverty measure, however,
is quite restrictive. If researchers or
policymakers wish to identify those
households in greatest need, this restric­tive
approach to identifying the poor
may be helpful.
Family Economics and Nutrition Review
References
1. Abdel-Ghany, M. and Schwenk, F.N. 1993. Functional forms of household expenditure
patterns in the United States. Journal of Consumer Studies and Home Economics 17:325-342.
2. Citro, C.F. and Michael, R.T. 1995. Measuring Poverty: A New Approach. National Academy
Press, Washington, DC.
3. Deaton, A. 1998. Getting prices right: What should be done? Journal of Economic
Perspectives 12(1 ):37-46.
4. Fisher, G.M. 1992. The development and history of the poverty thresholds. Social Security
Bulletin 55( 4 ):3-14.
5. Formby, J.P. 1996. Regional poverty and inequality. Proceedings of the 25th Annual
Conference of the Eastern Family Economics and Resource Management Association,
pp. 1-31.
6. Friedman, M. 1957. A Theory of the Consumption Function. Princeton University Press,
Princeton.
7. Garner, T.l. and Blanciforti, L.A. 1994. Household income reporting: An analysis of U.S.
Consumer Expenditure Survey data. Journal of Official Statistics 1 0( 1 ):69-91.
8. Greene, W.H. 1993. Econometric Analysis, 2nd Edition. Macmillan Publishing, New York.
9. Jorgenson, D.W. 1998. Did we lose the War on Poverty? Journal of Economic Perspectives
12( 1 ):19-96.
10. Lino, M. 1996. Income and spending of poor households with children. Family Economics
and Nutrition Review 9( 1 ): 2-13.
11. McGregor, P.P.L. and Borooah, V.K. 1992. Is low spending or low income a better indicator
of whether or not a household is poor: Some results from the 1985 Family Expenditure Survey.
Journal of Social Policy 21( 1):53-69.
12. Paulin, G.D. and Ferraro, D.L. 1994. Imputing income in the Consumer Expenditure
Survey. Monthly Labor Review 117(12):23-31.
13. Rogers, J.M. and Gray, M.B. 1994. CE data: Quintiles of income versus quintiles of outlays.
Monthly Labor Review 117( 12):32-37.
14. Slesnick, D.T. 1996. Consumption and poverty: How effective are in-kind transfers?
Economic Journali06(439):1521-I545.
15. Slesnick, D. T. 1993. Gaining ground: Poverty in the postwar United States. Journal of
Political Economy 101( 1): 1-38.
16. U.S. Department of Labor. 1994. Consumer Expenditure Survey: Interview Survey. U.S.
Department of Labor, Bureau of Labor Statistics.
1999 Vol. 12 No.2 25
Ron Jevning
Los Angeles International
University-Irvine
Mark Biedebach
California State University
RajenAnand
Center for Nutrition Policy
and Promotion
26
Commentary
Cruciferous Vegetables and Human
Breast Cancer: An Important
Interdisciplinary Hypothesis in
the Field of Diet and Cancer
Very early progress in cancer treatment and prevention was based primarily
on a basic understanding of genetic changes in genes at cellular and biochemical
levels. Today, however, an interdisciplinary approach from complementary
research tracks is possible in the understanding of cancer treatment and
prevention. Such an approach is particularly important for its potential to
increase our knowledge about diet and cancer because it may lead to
sounder dietary guidance. This interdisciplinary approach is well illustrated
by a hypothesis linking cruciferous vegetables to breast cancer prevention.
The hypothesis links indole-3-carbinol, a specific component of brassica
vegetables, such as broccoli or cauliflower, to a beneficial effect on human
breast cancer (estrogen metabolism). In addition to its value for preventing
human breast cancer, the biologic elements of the hypothesis have specific
implications for research on other cancers and for other diets.
Cancer: Trends, Complexity,
and Research
All have concerns about cancer. Because
of the incidence and devastating effects
of cancer, its burden of suffering and
death throughout the world is huge. In
English-speaking countries, the inci­dence
of cancer appears to be increasing
at an alarming rate (26). However,
because it takes several years to collect
and analyze sufficient data to establish
trends, the trend in English-speaking
countries is only probable. For example,
the incidence of prostate cancer increased
from 27,000 to 41 ,000 per year in the
8 years preceding 1991 (2). In 1992,
180,000 new cases of breast cancer
were reported in American women,
compared with 142,000 in 1989 (20).
There also seems to be a steady increase
in mortality from most cancers in recent
years, with the most rapid increase
occurring in steroid-related cancers
(see table). Both these projections­increased
incidence and increased
mortality-are supported by a recent
review on the topic (26).
Cancer is a very complex disorder;
hence, research on its cure and preven­tion
has had to use several approaches.
Early researchers tried almost exclu­sively
to understand cancer by studying
biochemical and genetic effects of such
cancer-causing emissions as X rays,
ultraviolet radiation, radioactive emana­tions,
and the effects of chemical agents
(5,9,11,24). However, later scientists
began to use statistical or epidemiological1
approaches to examine whether envi­ronmental
factors such as diet or life­style
may affect cancer risk (7,10,15).
1 "Epidemiological" refers to the study of
diseases within particular groups or populations.
Family Economics and Nutrition Review
Cancer deaths per year from the most deadly types
Percent
1992 1996 (estimated) increase of
Types total over
Male Female Total Male Female Total 4 years
Lung 91,400 54,500 145,900 94,400 64,300 158,700 8.8
Pancreas and colo-rectal 41,100 42,400 83,500 41,600 42,300 83,900 0.5
Steroid-involved
(Breast, ovary, and prostate) 34,200 56,500 90,700 41,400 59,100 100,500 10.8
Source: Boring, C. C., Squires, J.S., and Tong, T. 1992. Cancer statistics. Cancer Journal for Clinicians 42:19-35. Estimates are projections from 1990-1992
trends.
These latter approaches have led many
to conclude that diet has a likely role in
cancer. In particular, Doll and Peto (9)
in the United States and a group of
researchers in Sweden (5) believe that
the disease could be reduced by as
much as 35 percent by practical dietary
means.
One of the problems with these statistical
methods is that they do not provide
precise understanding of what about
the diet may be associated with change
of cancer risk (22,25). For example,
investigators are fairly certain that diet
change can reduce the risk of breast
cancer. But is fat the culprit in breast
cancer? Is the relatively higher concen­tration
of fruits and vegetables in im­proved
diets responsible? Complementing
these statistical approaches with bio­chemical
and genetic data is an indis­pensable
input into providing more
sound dietary guidance. In this article
we describe a specific theory about diet
and breast cancer that illustrates the ad­vantages
of such a combined approach
for practical dietary guidance.
1999 Vol. 12 No. 2
Diet-Estrogen Link to Breast
Cancer
In steroid-related cancers, the tissues
affected are those associated with
reproduction. Epidemiologically,
changes in incidence of these cancers
correlate directly with dietary change
( 12). In particular, the incidence of breast
cancer in English-speaking countries
is between 10 and 15 times the incidence
in poorer countries such as Thailand
or Ecuador, a fact attributed by some
researchers to difference in diet (7) (see
figure). However, a serious problem
with these correlational studies is that
we often do not know or control for
risk factors related to breast cancer­including
both low parity (small number
of offspring) and late age at first birth­other
than diet that operate in developed
countries. In this commentary, we
maintain that a better understanding
of the physiology of breast cancer can
help clarify what it is about the diei
that may affect breast cancer.
Early physiologically based research
has suggested a role for estrogen2 in
breast cancer (1,19). Much later, re­search
has suggested that diet probably
influenced the levels of blood estrogen
( 19,20,23 ). In 1996, Beatson noted that
removal of the ovaries containing the
estrogen-releasing cells was beneficial
in some cases of breast cancer ( 1 ). In
1990, Key et al. found that compared
with British women, rural Chinese
women had lower estrogen levels and
one-fifth the incidence of breast cancer
(15).
At the same time, strong biochemical
evidence links estrogen to cancer of
reproductive tissue. Estrogen activates
the parts of the chromosome (DNA)
that promotes cell division. We know,
however, that more than one form of
estrogen exists: estradiol, the form nor­mally
secreted, promotes cell division
in a well-controlled manner, while 16-
hydroxyestrone (C-16), another form
of estrogen, seems to promote cell divi­sion
in an uncontrolled manner that can
lead to cancer in affected tissue (23 ).
2Estrogen is a steroid hormone that acts during
the menstural cycle to prepare the uterine and
mammary tissues for possible pregnancy.
27
C-16 is one of the metabolites of normal
estrogen, from which it differs only by
the presence of a hydroxyl group on
the number 16 carbon atom; C-2, the
alternative estrogen metabolite, is a
"safe" (inactive) substance.
Finally, we know that women with a
family history of breast cancer have
elevated blood C-16 ( 4 ); and the anti­estrogenic
prescription drug 4-hydroxy
tamoxifen (or simply tamoxifen)
lowers blood C-16 and reduces both
the incidence of breast cancer and the
growth rate of existing breast cancer
cells (1 3,14,21 ).
A Biochemically and
Physiologically Based Breast
Cancer-Diet Hypothesis
In the 1970's, Wattenberg reported
that a diet of cruciferous (specifically
brassica) vegetables3 was associated
with an anti-breast cancer effect in
animals (24). Wattenberg also identified
a chemical compound in the vegetables,
indole-3-carbinol (i-3-C), as the main
contributor to their anti-cancer effect.
Later authors have reported decreased
formation of C-16 associated with in­creased
vegetable diets (5). Such findings
as these and those we discussed above
led Michnovicz to hypothesize that
cruciferous vegetables or purified i-3-C
may reduce breast cancer initiation or
the C-16/C-2 ratio by decreasing C-16
in the metabolism of estrogen ( 19,20).
While there is already some experimental
support for this hypothesis (3,19), larger
clinical studies of i-3-C are necessary
and are in progress (20).
3 Among these types of vegetables are cabbage,
broccoli, cauliflower, and Brussels sprouts.
28
Cancer death rate versus fat intake
Death rate
(per 100,000,
age adjusted)
30 ,-----------------------------------------~
25
20
15
10
5
• Netherlands
•Canada
• United States
•Australia
• Czechoslovakia
• Hungary
• Poland
• Yugoslavia
Thailand • • El ~alvador
0 ~------~--------_J--------~-1 -------_J
0 50 100 150 200
Grams of fat ingested per day
Importance of Breast-Diet
Hypothesis
Scientifically based references such as
the Food Guide Pyramid or the Dietary
Guidelines for Americans, published
by the U.S. Department of Agriculture,
exemplify bodies of knowledge relating
to dietary guidance that may change
with time. Therefore, on the basis of
much research (some of which we have
described), we know that Americans
should probably choose a diet low in
fat and cholesterol and one with plenty
of vegetables and grains. However,
much of our knowledge lacks specificity.
For example, are there particular diets
that may protect against cancer?
The possible link among crucifers,
estrogen, and breast cancer illustrates
how change in our knowledge will
probably occur. Therefore, even if the
details of this hypothesis prove wrong,
its epidemiological, physiological, and
biochemical basis has already stimu-lated
new research and discoveries.
To exemplify this, we first note that
net C-16 removal is mediated by repre­sentatives
of a particular enzyme group
known as "mixed function oxidases"
(MFO's) (24) and that i-3-C activates
these enzymes (8). MFO's have a
generalized tissue function of rendering
toxic substances harmless (including
many otherwise carcinogenic com­pounds).
Questions important for
cancer in general then have arisen that
include "what other substances than
i-3-C can activate the C-16 removing
MFO's? In what other tissues can
general MFO activity be increased?"
For example, it has been reported that
ascorbigen, a vitamin-C group com­pound
(6), activates MFO and that there
is marked synergy in MFO activation
by administration of both i-3-C and
ascorbigen (even though ascorbigen is
a nucleoside that itself contains i-3-C)
( 5, 16, 17). McDanell also reports MFO
activation or a synergistic effect of
joint i-3-C and ascorbigen on MFO
Family Economics and Nutrition Review
activity in a wide variety of tissues,
including small intestine, large intestine,
liver, and lung ( 17). We can then rea­sonably
ask this: "Are there cancer
prevention or treatment implications
of i-3-C or ascorbigen for tissues other
than breast tissue? Is MFO activation
operative in the reported anti-cancer
effect of a vegetable diet on a wide
variety of these tissues? Can benefits
be enhanced by combination i-3-C/
ascorbigen supplementation or
administration?"
General principles of the hypothesized
beneficial linkage between crucifers
and breast cancer may also be relevant
to other diets or dietary aspects. There­fore,
the central role of estrogen suggests
that a fiber-rich diet may be protective,
because dietary fiber increases removal
and decreases reabsorption of stool
estrogen. This biologic rationale com­plements
some epidemiologic evidence
that fiber is important, although epidemio­logic
assessment has been compromised
by scarcity of data on the fiber content
of individual foods (25 ). Similarly,
suggestive data, indicating protection
for a soy diet, have a biologic rationale
in the ability of soy isoflavones to
interfere with estrogen receptors ( 18).
Finally, selection for future evaluation
from the enormous number of phyto­chemical
possibilities can be guided,
in part, by knowledge of which chemicals
affect estrogen content or biochemistry.
The possible cruciferous vegetable
and estrogen linkage exemplifies how
more sound and more specific guidance
can result from combining pieces of
the dietary puzzle from a variety of
scientific disciplines. While we focused
on only breast cancer here, the possible
linkage between cruciferous vegetables
and estrogen may have far wider sig­nificance
for other diets and other cancers,
because the principles discussed here
are generally applicable.
1999 Vol. 12 No.2
References
1. Beatson, G.T. 1996. On the treatment of inoperable cases of carcinoma of the mamma:
Suggestions for a new method of treatment with illustrative cases. Lancet 2:104-107.
2. Boring, C. C., Squires, J.S., and Tong, T. 1992. Cancer statistics. Cancer Journal for
Clinicians 42: 19-35.
3. Bradfield, C.H. and Bjeldanes, L.F. 1984. Effect of dietary indole-3-carbinol on intestinal
and hepatic monooxygenase, glutathione S-transferase and epoxide hydrolase activities in
the rat. Food and Chemical Toxicology 22:977-987.
4. Bradlow, H.L. and Michnovicz, J.J. 1989. A new approach to the prevention of breast
cancer. Proceedings of the Royal Society, Edinburgh 95B:77-86.
5. Byers, T. and Graham, S. 1984. The epidemiology of diet and cancer. Advances in Cancer
Research 41 : 1.
6. Cameron, E. and Pauling, L. 1979. Ascorbic acid and cancer: A review. Cancer Research
39:663.
7. Cohen, L.A. 1987. Diet and cancer. Scientific American 257:42-50.
8. Dash wood, R.H., Arbogast, D.N., Fong, A.T., et al. 1988. Mechanism of anticarcinogens
by indole-3-carbinol: Detailed in vivo DNA binding dose-response studies after dietary
administration with aflatoxin B 1. Carcinogenesis 9:427-432.
9. Doll, R. and Peto, R. 1981 . The causes of cancer: Quantitative estimates of avoidable risks
of cancer in the United States today. Journal of the National Cancer Institute 66: 1191-1308.
I 0. Frisch, R.E., Wyshak, G., Witschi, J., et al. 1987. Lower lifetime occurrences of breast
cancer and cancers of the reproductive system among former college athletes. International
Journal of Fertility 32:217-225.
11. Gerhardsson, J.D. and Donahue, L. 1988. Aflatoxin, a human carcinogen: Determination in
foods and biological samples by monoclonal antibody affinity chromatography. Journal of
the Association of Official Analytical Chemists 71 :861-867.
12. Haensze1, L. and Kurihara, M. 1968. Studies of Japanese migrants. I. Mortality from
cancer and other diseases among Japanese in the United States. Journal of the National
Cancer Institute 40:43-49.
13. Han, X. and Liehr, J.G. 1992. Induction of covalent DNA adducts in rodents by tamoxifen.
Cancer Research 52:1360-1363.
29
14. Harris, J.R., Lippman, M.E., Veronesi, U. 1992. New England Journal of Medicine
327:389-398.
15. Key, T.J.A., Chen, J., Wang,D.Y., et al. 1992. Sex hormones in women in rural China
and in Britain. British Journal of Cancer 62:631-636.
16. Kutacek, M., ProchazkaZ., and Valenta, M. 1962. The metabolism of glucobrassicine
and other indole derivatives in brassica, in naturally occurring goitrogens, thyroid function.
Symposium, Smolensk, Czechoslovakia pp. 49-56.
17. McDanell, R., McLean, A.E.M., Hanley, A.B., et al. 1987. Differential induction of
mixed-function oxidase (MFO) activity in rat liver and intestine by diets containing processed
cabbage: Correlation with cabbage levels of glucosinolates and glucosinolate hydrolysis
products. Food and Chemical Toxicology 25:363-368.
18. Messina, M. and Barnes, S. 1991 . The role of soy products in reducing risk of cancer.
Journal of the National Cancer Institute 83:541-546.
19. Michnovicz, J.J., Adlercreutz, H., and Bradlow, H.L. 1997. Changes in levels of urinary
estrogen metabolite after oral indole-3-carbinol treatment in humans. Journal of the National
Cancer Institute 89:718-723.
20. Michnovicz, J.J. and Klein, D.S. 1994. How to Reduce Your Risk of Breast Cancer.
Warner Books, New York.
21 . Nayfield, S.G., Karp, J.E., Ford, L.G., et al. 1991 . Potential role of tamoxifen in prevention
of breast cancer. Journal of the National Cancer Institute 83:1450-1459.
22. Steinmetz, K.A. and Potter, J.D. 1991. Vegetables, fruit and cancer. I. Epidemiology.
Cancer Causes and Control2:325-351.
23. Telang, N.T., et al. 1997. Inhibition of proliferation and modulation of estradiol metabolism;
novel mechanisms for breast cancer prevention by the phytochemical indole-3-carbinol.
Proceedings of the Society of Experimental Biology and Medicine 215:246-254.
24. Wattenberg, L.W. 1997. An overview of chemoprevention; current status and future prospects.
Proceedings of the Society of Experimental Biology and Medicine 216: 133-155.
25. Willett, W.C. and Hunter, D.J. 1992. Dietary fat and fiber in relation to breast cancer.
Journal of the American Medical Association 268:2034-2044.
26. World Cancer Research Fund/ American Institute for Cancer Research. 1997. Food, Nutrition
& the Prevention of Cancer; a global perspective. Washington, DC.
30 Family Economics and Nutrition Review
Shanthy A. Bowman
U.S. Department of Agriculture
Agricultural Research Service
"Pooh, do you want honey
in your tea?" "Yes, Piglet,
but without the tea."
-Walt Disney Winnie the Pooh
video series.
1999 Vol. 12 No.2
Research Briefs
Diets of Individuals Based
on Energy Intakes
From Added Sugars
Data from the U.S. Department of
Agriculture's (USDA) Nationwide Food
Consumption Survey (NFCS) and the
Continuing Survey of Food Intakes
by Individuals (CSFII) show a steady
increase in people's total energy intake
since 1987 (9,10,12). Diet quality as
measured by the Healthy Eating Index
showed that, from 1989 to 1996, while
the intakes of grain products increased
appreciably, those of milk decreased ( 1 ).
Harnak et al. (3), in a study using CSFII
1994 data, reported that a high level of
soft drink consumption by children and
adolescents was associated with low
intakes of milk and fruit juices and
with low intakes of several nutrients:
such as calcium, phosphorus, riboflavin,
vitamin A, folate, and vitamin C. Non­diet
soft drinks; fruit drinks; and foods
such as cakes, cookies, and pies, placed
under grain products in the CSFII, are
high contributors of added sugars in
the American diet.
According to per capita data from the
U.S. food supply, consumption of added
sugars in 1997 was 53 teaspoons per day,
reflecting a 28-percent increase from
1982 (7). Added sugars are generally
considered "empty calories," because
added sugars are good sources of energy
and often are poor sources of micro­nutrients.
This study examines the intakes
of food groups and nutrients by indi­viduals
grouped by the caloric contri­bution
of added sugars to their diet.
The study also attempts to determine
whether high intakes of added sugars
displace essential nutrients or nutrient­dense
foods in the individual's diet.
Method
CSFII Definition of Added Sugars
Added sugars include all sugars used
as ingredients in processed and pre­pared
foods such as breads, cakes, soft
drinks, jam, and ice cream, and sugars
eaten separately or added to foods at
the table ( 10). Specifically, added sugars
include white sugar, brown sugar, raw
sugar, com syrup, com-syrup solids,
high-fructose com syrup, malt syrup,
maple syrup, pancake syrup, fructose
sweetener, liquid fructose, honey
molasses, anhydrous dextrose, and
crystal dextrose. Added sugars do not
include naturally occurring sugars such
as lactose in milk or the fructose in
fruits.
Data Source
Data from USDA's 1994-96 CSFII,
a nationally representative food con­sumption
survey, were used for this
study (8). The dietary data were col­lected
on 2 nonconsecutive days (3 to
10 days apart). A multiple-pass approach
was used to collect two interviewer­administered
24-hour recalls. Individuals
2 years old and over who were selected
for this study had a complete food
intake record on day 1 of the survey .1
1overall, the response rate for day I was 80
percent and included 15,016 individuals2 years
old and over (8).
31
Analysis revealed some extreme values
for the day's total energy. Therefore,
the top (more than 5,200 kcals) and
bottom (less than 490 kcals) 1 percent
of the individuals were excluded from
the analysis. Also, excluded from the
analysis were three individuals who
had energy intakes entirely from added
sugars. The analysis included 14,709
individuals.
The individuals were divided into three
groups based on the percentage of
calories consumed from added sugars.
Group 1 (N=5,058) had less than 10
percent of its total calories from added
sugars; group 2 (N=4,488), between
10 and 18 percent; and group 3
(N=5, 158), more than 18 percent of
its total calories from added sugars.
Data Analysis
Day-1 full sample weights were used
to represent the population under study.
SUDAAN (release 7.5.1, Research
Triangle Institute) was used to compare
the three groups' mean intakes of food
groups, nutrients, and energy. Three
pairwise comparisons of the means
were made, and linear contrasts were
used to separate the means. A prob­ability
level of0.0125 was used to keep
the total experimental error rate low,
and SAS software package (release
6.12, SAS Institute, Cary, NC) was
used to compute all the other
estimations.
Results
Group 1 had the lowest intakes of energy
and added sugars among the three groups
(table 1). Group 1 consumed 1,860
kcal and 26 grams of sugar: 180 to 189
kcal and 45 to 111 grams less than that
consumed by the other groups. There
were no significant differences in total
fat and saturated fat intakes of group 1
32
Table 1. Mean1 intakes of energy, macronutrients, and percentage of
calories in a day by individuals 2 years and over, by percentage of
calories from added sugars
Calories from added sugars
Less than 10% 10 to 18% Above 18%
Energy and nutrients (Group I) (Group 2) (Group 3)
Sample 5,058 4,488 5,158
Mean S.E.2 Mean S.E.2 Mean S.E.2
Energy (kcal) 1860a 15.0 2040b 18.1 2049b 17.2
Total fat (g) 73a 0.9 78b 0.1 70a 0.8
Saturated fat (g) 24a 0.3 27b 0.4 25a 0.3
Carbohydrate (g) 211 a 1.7 256b 2.2 292c 2.4
Protein (g) 81a 0.7 78b 0.7 66c 0.7
Dietary fiber (g) 17a 0.2 16a 0.2 13b 0.2
Added sugars (g) 263 0.3 71b 0.7 137c 1.3
Percent of calories 35.3 34.4 30.7
from total fat (% )3
Percent of calories 5.6 13.9 26.7
from added sugars (%l
1 Means with identical superscripts are not significantly different from each other at p < 0.0125.
2Standard error of the mean.
3No statistical test of significance was done.
Note: Linear contrasts were used to separate the means.
Source: USDA 's Continuing Su111ey ofF ood Intakes by Individuals 1994-96, Day-1 data.
and group 3, but their fat intakes were
lower than those of group 2. The diets
of all three groups, however, had more
than 30 percent of calories from total
fat.
Group 1, consuming less than 10 percent
of calories from added sugars, had much
higher intakes of protein and dietary
fiber than did group 3, which consumed
more than 18 percent of calories from
added sugars. Although group 1 had a
diet with apparently less carbohydrate
than did the other two groups, when
the added sugars were subtracted from
total carbohydrate, the amount of
carbohydrate without the added sugar
was the same as that of group 2, and
much higher than that of group 3. That
is, compared with group 3, group 1 had
a diet higher in carbohydrate without
the added sugars.
Group 3, having consumed more than
18 percent of calories from added sugars,
had the lowest mean absolute intakes
of all the micro nutrients, especially
vitamin A, vitamin C, folate, vitamin
Bt2, calcium, phosphorus, magnesium,
and iron (table 2). Group 1 and group 2
had similar intakes of most micro­nutrients
in absolute amounts; the
Family Economics and Nutrition Review
-Table 2. Mean1 intakes ofmicronutrients in a day by individuals 2 years
and over, by percentage of calories from added sugars
Calories from added sugars
Nutrients
Sample
Thiamin (mg)
Riboflavin (mg)
Vitamin A (RE)
Vitamin E (mg)
Vitamin C (mg)
Niacin (mg)
Vitamin B6 (mg)
Folate (meg)
Vitamin B12 (meg)
Calcium (mg)
Phosphorus (mg)
Magnesium (mg)
Iron (mg)
Zinc (mg)
Copper (mg)
Less than 1 0%
(Group 1)
5,058
Mean S.E.2
1.6a 0.01
1.9a 0.02
1080a 23.0
8.3a 0.13
106a 2.2
23a 0.2
1.9a 0.02
275a 3.6
5.4a 0.18
788a 8.1
1251a 9.4
285a 2.4
15.6a 0.14
11.5a 0.15
1.2a 0.01
10% to 18% Above 18%
(Group 2) (Group 3)
4,488 5,158
Mean S.E.2 Mean S.E.2
1.7a 0.02 1.5b 0.02
2.0b 0.02 1.8c 0.02
1031a 26.8 850b 20.2
8.4a 0.15 7.lb 0.12
lOla 1.8 90b 1.5
23a 0.3 20b 0.3
1.9a 0.02 1.6b 0.02
272a 3.8 228b 3.6
5.2a 0.23 4.3b 0.13
838b 10.1 745c 11.0
1277a 12.5 1130b 11.4
277a 3.1 233b 2.6
16.1a 0.19 14.1b 0.18
11.6a 0.15 lO.lb 0.13
1.2a 0.01 l.lb 0.01
1 Means with identical superscripts are not significantly different from each other at p < 0.0125.
2Standard error of the mean.
Note: Linear contrasts were used to separate the means.
Source: USDA's Continuing Survey ofF ood Intakes by Individuals 1994-96, Day-1 data.
exceptions were riboflavin and calcium.
Group 3 also had the lowest intakes of
all the micronutrients as measured by
percentages of 1989 Recommended
Dietary Allowances (RDA) (table 3).
All three groups had mean intakes less
than 100 percent of the RDA for vitamin
E, calcium, and zinc (6). In addition,
group 3 had mean intakes less than 100
for vitamin B6 and magnesium. A remark­ably
lower percentage of individuals in
group 3 met their RDA for many micro­nutrients
(table 4). However, more or
1999 Vol. 12 No. 2
less similar percentages of individuals
in groups 1 and 2 met the RDA for the
micronurients. Whereas about one-fourth
of the individuals with a low intake of
added sugars (group 1) had energy
intakes that equaled or exceeded the
Recommended Energy Allowances
(REA), about one-third each of the
individuals with moderate or high
intakes of added sugars did so.
In addition, group 3 (more than 18
percent of calories from added sugars)
Group 3 had the
lowest intakes of all
the micronutrients.
33
had the lowest intakes of many food
groups: Grain; Fruit; Vegetables; and
Meat, Poultry, and Fish (table 5). Indi­viduals
with low intakes of added sugars
included more fruits; vegetables; and
meat, poultry, and fish in their diet,
compared with food intakes of the other
two groups. Dairy intake was the same
for the groups with low (group 1) or
high intake (group 3) of added sugars.
Analysis of mean intakes of selected
food subgroups shows that group 3
consumed less citrus and noncitrus
fruit juices and total fluid milk than
did the other two groups (table 6). Also,
compared with the other groups, group
3 had the highest intakes of regular
fruit drinks, punches, and ades; regular
carbonated soft drinks; cakes, cookies,
and grain-based pastries; milk desserts;
and candies. Group 2 had the second
highest intakes, and group 1 had the
least intakes of these food subgroups.
The increase was more than tenfold
between group 1 and group 3 for regular
fruit drinks, punches, and ades; and
regular carbonated soft drinks.
Additional analysis showed that among
males, 34 percent were in group 1; 30
percent, group 2; and 36 percent in
group 3. Similar percentages of females
were in each group: 33 percent were
in group 1; 30 percent, group 2; and
37 percent were in group 3. Forty-four
percent of African Americans, com­pared
with 33 percent of Caucasians,
were in group 3. Among the individuals
from households with income less than
300 percent of poverty, about 40 percent
were in group 3, compared with less
than one-third who were in group 1.
When household income levels were
at or above 300 percent of poverty,
individuals were about as likely to be
in group 3 as in group 1: 34 and 35
percent, respectively.
34
Table 3. Meanl intakes of energy and micronutrients as percentage of
1989 Recommended Dietary Allowances (RDA) in a day by individmls
2 years and over, by percentage of calories from added sugars
Calories from added sugars
Energy and nutrients Less than 10% 10% to 18% Above 18%
(%RDA) (Group 1) (Group 2) (Group 3)
Sample 5,058 4,488 5,158
Mean S.E.2 Mean S.E.2 Mean S.E.2
Energy 81a 0.5 89b 0.8 88b 0.8
Vitamin A (RE) 128a 2.8 125a 3.0 103b 2.4
Vitamin E 95a 1.4 97a 1.6 82b 1.5
Vitamin C 185a 3.9 179a 3.0 163b 2.8
Thiamin 139a 1.2 143a 1.5 124b 1.5
Riboflavin 141a 1.1 148b 1.8 130c 1.7
Niacin 150a 1.1 150a 1.7 129b 1.4
Vitamin B6 111a 1.1 112a 1.4 94b 1.1
Folate 167a 2.5 175a 2.5 149b 2.5
Vitamin B12 289a 9.6 285a 11.8 237b 7.0
Calcium 93a 1.0 97a 1.1 83b 1.2
Phosphorus 149a 1.1 148a 1.4 126b 1.4
Magnesium 104a 0.9 105a 1.4 89b 1.1
Iron 142a 1.4 145a 1.8 124b 1.6
Zinc 87a 1.1 89a 1.1 78b 1.0
1 Means with identical superscripts are not significantly different from each other at p < 0.0125.
2standard error of the mean.
Note: Linear contrasts were used to separate the means.
Source: USDA's Continuing Survey of Food Intakes by Individuals 1994-96, Day-/ data.
Independent of gender, the percentage
of individuals with more than 18 percent
of calories from added sugars (group 3)
increased from the childhood years to
the teen years and declined in the adult
years (table 7). About one-third of
children 2 to 5 years old, and one-half
of children 6 to 11 years old were in
group 3.
Discussion and Conclusion
High intake of added sugars had a dilu­tion
effect on many essential micro­nutrients-
especially vitamin A, vitamin
B12, folate, magnesium, and iron-in
the diet of Americans 2 years old and
over. Individuals who consumed more
than 18 percent of calories from added
sugars had low intakes of all the five
food groups.
Family Economics and Nutrition Review
Table 4. Percentage of individuals 2 years and over meeting 100 percent
of 1989 Recommended Dietary Allowances (RDA) for selected nutrients
and energy in a day, by percentage of calories from added sugars
Calories from added sugars
Energy and nutrients
(% RDA)
Less than 10% 10% to 18% Above 18%
(Group 1) (Group 2) (Group 3)
Sample 5,058
Energy 24
Protein 78
Vitamin A (RE) 42
Vitamin E 34
Vitamin C 59
Thiamin 68
Riboflavin 68
Niacin 74
Vitamin B6 49
Folate 66
Vitamin B12 75
Calcium 37
Phosphorus 73
Magnesium 42
Iron 62
Zinc 29
Note: SAS analysis of weighted data.
4,488
Percent
32
80
44
35
61
71
72
74
50
67
78
39
75
43
64
32
5,158
32
70
34
26
52
59
61
62
37
55
73
30
62
31
52
23
Source: USDA 's Continuing Survey ofF ood Intakes by Individuals 1994-96, Day-1 data.
And compared with other groups,
group 3 had the lowest intakes for all
the micronutrients. Thus group 3 had
the least nutrient-dense diet. Adequate
intake of rnicronutrients has implications
for long-term well-being. A high per­centage
of the adult skeleton is formed
during adolescence (5). Thus adequate
intake of calcium during childhood and
adolescence is essential. Also increased
risk of osteoporosis is associated with
1999 Vol. 12 No.2
low bone density, which results from
inadequate intakes of calcium during
the growing years (2).
With the lowest intakes of both energy
and added sugars, group 1 did not have
a lower percentage of calories from
added sugars by eating more of other
energy-giving nutrients (thereby in­creasing
the denominator) but by
controlling the intake of added sugars.
High intake of added
sugars had a dilution
effect on many
essential micro­nutrients
.. .in the diet
of Americans 2 years
old and over.
35
Although the individuals in group 1
had more than 30 percent of total calories
from fat, their absolute mean intakes of
total fat and saturated fat were similar
to those of group 3 and less than those
of group 2. Though only one-fourth of
group I met their energy requirements,
in the cases of micronutrients, the per­centage
of individuals meeting the
recommended nutrient levels were
comparable to that of group 2.
Compared with the others, group 3 had
a high-energy and a relatively lower
fat diet. A study by Kennedy et al. ( 4 ),
using day- I data from CSFII I995,
showed that children 6 to 18 years old
and females I9 to 50 years old whose
diets had less than 30 percent of calories
from fat had a higher intake of total
sugars and sweets and total beverages
(excluding milk and fruit juices). The
same study also showed that all adults
ages 19 to 50 with diets less than 30
percent of calories from fat and whose
diets did not include any fat-modified,
lean or lower fat food products had a
high intake of regular carbonated
beverages.
The study showed that children were
more likely to have a diet high in added
sugars. Adults over 40 years were likely
to have lower added sugar intakes.
Within the same age group, the gender
did not seem to affect the percentage
of caloric contribution of added sugars.
Group 3 had high consumption of
beverages that are very low in nutrients
and high in energy. Because of the
increasing prevalence of obesity,
consumers will be benefitted by limiting
intake of "empty" calories, especially
during childhood and adolescence.
36
Table 5. Meant intakes of Pyramid food group servings in a day by
individuals 2 years and over, by percentage of calories from added sugars
Food groups
Sample
Grain
Fruit
Vegetable
Dairy
Meat, poultry, and
fish (ounces)
Calories from added sugars
Less than 10% 10% to 18% Above 18%
(Group 1) (Group 2) (Group 3)
Mean
6.7a
l.8a
3.7a
l.5a
4.6a
5,058
S.£.2
0.08
0.04
0.06
0.02
0.06
4,48S
Number of servings
Mean S.£.2
7.1b 0.07
l.6b 0.04
3.5b 0.05
I.6b 0.03
4.3b 0.07
Mean
6.3c
l.2c
2.9c
l.4a
3f
5,158
S.£.2
0.07
0.04
0.05
0.03
0.07
I Means with identical superscripts are not significantly different from each other at p < 0.0125.
2standard error of the mean.
Note: Linear contrasts were used to separate the means.
Source: USDA 's Continuing Survey of Food Intakes by Individuals 1994-96, Day-1 data.
Table 6. Mean1 intakes of selected food subgroups in a day by individuals
2 years and over, by percentage of calories from added sugars
Calories from added sugars
Food subgroups Less than I 0% 10% to 18% Above 18%
(Group 1) (Group 2) (Group 3)
Sample 5,058 4,488 5,158
Grams
Mean S.£ .2 Mean S.£.2 Mean S.£.2
Citrus juices 79a 4.1 63b 3.8 42c 2.1
Noncitrus fruit juices 33a 2.1 28a 1.6 16b 1.3
and nectars
Total fluid milk 201a 5.6 205a 5.0 160b 3.8
Regular fruit drinks, 12a 1.2 67b 3.0 149c 5.7
punches, and ades
Regular carbonated 32a 2.1 176b 5.5 515c 14.0
soft drinks
Cakes, cookies, and 18a 0.8 43b 1.6 53c 1.8
grain-based pastries
Milk desserts lla 0.7 27b 1.1 41c 1.8
Candies 2a 0.2 6b 0.4 13c 0.7
1 Means with identical superscripts are not significantly different from each other at p < 0.0125.
2standard error of the mean.
Note: Linear contrasts were used to separate the means.
Source: USDA 's Continuing Survey ofF ood Intakes by Individuals 1994-96, Day-1 data.
Family Economics and Nutrition Review
Table 7. Percentage of individuals, by age-gender and by percentage
of calories from added sugars
Calories from added sugars
Age (years)- Less than 10% 10% to 18% Above 18%
gender group (Group 1) (Group 2) (Group 3)
Sample 5,058 4,488 5,158
Percent
All individuals 34 30 36
Child 2-5 29 36 35
Child 6-11 30 31 49
Male 12-18 16 28 56
Female 12-18 17 30 53
Male 19-40 32 31 37
Female 19-40 32 28 40
Male 41 and over 45 30 25
Female 41 and over 43 30 27
Note: SAS analysis of weighted data.
Source: USDA 's Continuing Survey ofF ood Intakes by Individuals 1994-96, Day-/ data.
African Americans and low-income
individuals were more likely than their
counterparts to have high intakes of
added sugars. It is possible that foods
high in added sugars were less expen­sive
energy sources for at least some
of the individuals in the low-income
group. Income could play a role in the
choice of foods because higher con­sumption
of expensive foods such as
fruits and vegetables was associated
with diets where added sugars were
low.
When the total fat intake meets the
recommendations of the Dietary Guide­lines
for Americans (13 ), the Food Guide
Pyramid suggested levels of added sugars
are 6, 12, and 18 teaspoons (24, 48,
and 72 grams) per 1,600, 2,200, and
2,800 calories of total energy per day,
respectively ( 11 ).The mean intake of
1999 Vol. 12 No.2
added sugars for group 3 was 137
grams, which is much higher than
these recommended levels.
Data on food disappearance (in the food
supply) show that more than three­quarters
of the refined and processed
sugars reach the consumer through
food and beverage industries, and less
than one-fourth of the amount produced
is brought directly into the home ( 1 ).
It is important for consumers to recog­nize
that they get large amounts of
added sugars through processed foods
and beverages. Additional analyses of
data from day 1 of the CSFII 1994-96
show that individuals 2 years old and
over (N=15,016) consume 20.5 tea­spoons
(82 grams) of added sugars
daily. The top five sources of added
sugars and their mean contribution to
the daily intakes of added sugars in the
diet are carbonated soft drinks (27
grams); cakes, cookies, pies, sweet
rolls, and other grain-based pastries
(11 grams); fruit drinks (excludes fruit
juices), punches, and ades (8 grams);
dairy desserts (4 grams); and all types
of candies (4 grams).
Food labels contain information on
total sugars per serving but do not
distinguish between sugars naturally
present in foods and added sugars.
Better information on the food label
is needed to help consumers make
informed choices regarding added
sugars.
Acknowledgment
The author wishes to thank the reviewers
for their helpful suggestions.
37
References
I. Bowman, S.A., Lino, M., Gerrior, S.A., and Basiotis, P.P. 1998. The Healthy Eating Index:
1994-96. Family Economics and Nutrition Review 11(3):2-14.
2. Food and Nutrition Board, Institute of Medicine. 1998. Dietary Reference Intakes. Calcium,
Phosphorus, Magnesium, Vitamin D, and Fluoride. National Academy Press, Washington, DC.
3. Harnack, L., Stang, J., and Story, M. 1999. Soft drink consumption among US children
and adolescents: Nutritional consequences. Journal of the American Dietetic Association
99(4):436-441.
4. Kennedy, E.T., Bowman, S.A., and Powell, R. 1999. Dietary-fat intake in the U.S. population.
Journal of the American College of Nutrition 18( 3 ):207 -212.
5. National Institute of Health Consensus Development Panel on Optimal Calcium Intake.
1994. Journal of the American Medical Association 272:1942-1948.
6. National Research Council, Subcommittee on the Tenth Edition of the RDAs, Food and
Nutrition Board. 1989. Recommended Dietary Allowances (10th ed.). National Academy
Press, Washington, DC.
7. Putnam, J. and Gerrior, S. 1999. Trends in the U.S. Food Supply. In E. Frazao (Ed.) America's
Eating Habits: Changes and Consequences (pp. 133-160). U.S. Department of Agriculture,
Agriculture Information Bulletin No. 750.
8. Tippett, K.S. and Cypel, Y.S. (Eds.). 1998 (May). Design and Operation: The Continuing
Survey of Food Intakes by Individuals and the Diet and Health Knowledge Survey, 1994-96.
U.S. Department of Agriculture, Agricultural Research Service. NFS Report No. 91-1.
9. U.S. Department of Agriculture, Agricultural Research Service. 1995. Food and Nutrient
Intakes by Individuals in the United States, 1 Day, 1989-91. Nationwide Food Surveys. Report
No. 91-2.
10. U.S. Department of Agriculture, Agricultural Research Service. 1998. Food and Nutrient
Intakes by Individuals in the United States, by Sex and Age, 1994-96. Nationwide Food Surveys.
Report No. 96-2.
11. U.S. Department of Agriculture, Human Nutrition Information Service. 1992. The Food
Guide Pyramid. Home and Garden Bulletin No. 252.
12. U.S. Department of Agriculture, Human Nutrition Information Service. 1993. Food and
Nutrient Intakes by Individuals in the United States, 1 Day, 1987-88. Nationwide Food
Consumption Survey 1987-88, NFCS Report No. 87-1 .1.
13. U.S. Department of Agriculture and U.S. Department of Health and Human Services. 1995.
Nutrition and Your Health: Dietary Guidelines for Americans (4th ed.). U.S. Department of
Agriculture. Home and Garden Bulletin No. 232.
38 Family Economics and Nutrition Review
Mark Lino
Julia M. Dinkins
Lisa Bente
Center for Nutrition Policy
and Promotion
A version of this research brief, along
with other articles examining the use
of food stamps to purchase dietary
supplements, appears in "The Use
of Food Stamps to Purchase Vitamin
and Mineral Supplements" (Food and
Nutrition Service, U.S. Department of
Agriculture, September 1999).
1999 Vol. 12 No.2
Household Expenditures
on Vitamins and Minerals
by Income Level
The Federal Food Stamp Program
provides a nutritional safety net for low­income
households by giving eligible
individuals allotments that may be used
to purchase food. These allotments are
based on the Thrifty Food Plan, a minimal
cost of a nutritious diet in the United
States. Although most foods can be
purchased with the allotments, dietary
supplements (vitamins, minerals, and
other nutritional supplements, such as
herbal products and amino acids) have
been excluded.
The recent welfare reform act (Personal
Responsibility and Work Opportunity
Reconciliation Act of 1996) required the
U.S. Department of Agriculture (USDA)
to conduct a " ... study of the use of food
stamps to purchase vitamins and minerals"
(Section 855). One specific request was
a study on " ... the purchasing habits of
low-income populations with regards
to vitamins and minerals." To address
these purchasing habits, this paper
examines low-income households'
expenditures on vitamins and minerals
and compares their expenditures with
those of non low-income households.
Research on dietary supplements has
focused primarily on use. One study
found that in 1992, 46 percent of the
U.S. population reported taking a vitamin
or mineral supplement in the past year
(9). An earlier study found that in 1987,
23 percent of the population reported
taking a daily vitamin or mineral supple­ment
( 13 ). Characteristics associated
with vitamin/mineral use include being
female (3,8,12,13), being White (3,5,13 ),
having a higher education (3,5,8), having
a higher income (3,5,8), and being older
(5,8,13). In addition, residing in the
West (3,12), consuming more fruits and
vegetables (6), playing a sport ( 10), and
having some health problems (2) were
associated with vitamin/mineral use.
Multivitamins, vitamin C, calcium, and
iron were the most commonly consumed
dietary supplements (7, 12,13 ).
Although there has been considerable
research on the use of dietary supplements,
almost none has focused on people's
expenditures and other purchasing habits
regarding these supplements. According
to industry estimates, total retail sales
in 1992 were $2.7 billion for vitamins,
$0.5 billion for minerals, and $0.5 billion
for other nutritional supplements, for a
total of $3.7 billion (4). Multivitamins
and vitamin C supplements account for
the largest percentage of these sales ( 4 ).
According to Applied Biometrics ( 1 ),
people purchase dietary supplements
most often at drugstores, and the main
reasons they report taking supplements
are to prevent disease and increase
energy.
One unused source of information
regarding expenditures on dietary
supplements is the Diary component
39
of the Consumer Expenditure Survey
(CE). These data were used in this study
to examine household purchases of
dietary supplements.
Data and Sample
The Diary component of the CE, con­ducted
by the Bureau of the Census for
the Bureau of Labor Statistics (BLS), is
an ongoing survey that collects data on
food and other selected expenditures,
income, and major sociodernographic
characteristics of consumer units. A
consumer unit consists of either (1) all
members of a particular household who
are related by blood, marriage, adoption,
or other legal arrangements; (2) two or
more people living together who pool
their incomes to make joint expenditure
decisions; or (3) a person living alone or
sharing a household with others or living
as a roomer in a private home or lodging
house or in permanent living quarters in
a hotel or motel, but who is financially
independent.
A national sample of consumer units,
representing the civilian noninstitution­alized
population, was selected and
asked to keep an expenditure Diary,
which covers two consecutive 1-week
periods. Every year the CE surveys
about 5,000 different consumer units
throughout the year. Each week of
diaries is deemed an independent
sample by BLS.
For this study, data from the 1994 CE
Diary were used. The !-week diaries
were linked so that information on a
consumer unit's food and other selected
expenditures could be obtained for a
2-week period. This was done because
it is unlikely that people purchase dietary
supplements on a weekly basis. Only
units that were complete income reporters
and contained only one household in the
40
housing unit were included. Complete
income reporters provide values for
major sources of income, such as wages
and salary, interest and dividends, and
Social Security. Consumer units with
one household (the two terms will be
used interchangeably from this point
on) were included to avoid confusion
over which household made the purchase.
Low-income households were .then
selected from the data set. Low income
was defined as having before-tax income
less than or equal to 130 percent of the
poverty threshold for a respective house­hold
size. This definition is used to
determine eligibility for food stamps.
The final sample consisted of 833 low­income
households. To place the expendi­tures
and other characteristics of these
households in perspective, USDA
researchers also selected a random
sample of 833 non low-income house­holds.
(Non low-income households were
defined as those with before-tax income
above 130 percent of the poverty threshold
for a respective household size.)
The CE public-use tape contains infor­mation
on total over-the-counter drug
purchases of households; 216 of the low­income
households and 305 of the non
low-income households had over-the­counter
drug purchases. The individualized
expenditures constituting these over-the­counter
drug purchases, such as expendi­tures
on aspirin, cough medicine, and
vitamins or minerals, are recorded in the
actual CE diaries but are not reported on
the public-use tapes. To obtain expendi­tures
on vitamins or minerals, a USDA
team of researchers examined the actual
diaries of the 521 (low-income and non
low-income) households reporting ex­penditures
in the over-the-counter drug
category. These diaries are located at
BLS; working at BLS, the USDA team
used identification numbers to match
data on the public-use tape and the diaries.
In the diaries, consumer units recorded
purchases of vitamins or minerals by
type (e.g., vitamin Cor calcium), brand
name (e.g., One-A-Day or Centrum), or
simply as "vitamin or mineral." The
respondent chose how to record these
purchases. Because researchers could
not group these purchases (with a rea­sonable
degree of accuracy) by type
of vitamin or mineral, all purchases of
vitamins and minerals were totaled.
Other nutritional supplements (e.g.,
amino acids and herbs) were grouped
under "other" food. BLS provided a list
of all households with such expenditures
in the 1994 CE, and researchers examined
these diaries for purchases of other nutri­tional
supplements. Few households
reported expenditures on these other
dietary supplements, and almost none
were in the sample of 1,666low~income
and non low-income households. Expendi­tures
on these other dietary supplements
were, therefore, not examined in this
study. Some of these other nutritional
supplements could be listed simply as
"other" food, so the actual percentage
of people purchasing them is higher.
Consequently, these cases could not be
identified.
Results
The characteristics of the low-income
sample (table 1) are consistent with Census
findings of the low-income population
( 14). Most heads or co-heads oflow-income
households 1 were not married, had a high
school diploma or less, and were either
under 30 or over 59 years old. Most low­income
households reported not receiving
food stamps in the past year. This may
1The head or co-head was defined as the person
who owns or rents the home; in cases of joint
ownership or renting status, the head or co-head
is decided arbitrarily.
Family Economics and Nutrition Review
Table 1. Characteristics of households, by income, 1994
Characteristics Low-income Non low-income
Before-tax income*
Weekly food expense*
Household size
Age (years)*1
Less than 30
30-39
40-49
50-59
60 and over
Education*1
Less than high school
High school diploma
Some college
College degree
Race*1
White
Non-White
Family type*
Husband-wife with children
Husband-wife without children
Single-parent with children
Single
Other2
Housing tenure*
Own
Rent
Food stamp receipt*
Receive
Do not receive
Region*
Urban
Northeast
South
Midwest
West
Rural
households
n =833
$8,780
$54
2.6
21
19
14
9
37
41
29
22
8
77
23
20
9
17
36
18
42
58
31
69
16
29
23
18
14
households
n = 833
Mean
$45,560
$87
2.6
Percent
14
24
24
14
24
13
31
24
32
89
11
33
23
5
24
15
72
28
2
98
20
26
20
21
13
*Significant difference at .05 level.
1 Age, education, and race are for the reference person or household head or co-head, who is the person
who owns or rents the home; when there is joint ownership or renting status, the head or co-head is
decided arbitrarily.
20ther consists of hu