Dietary
M Popkin,
in older Americans,
Pamela
ABSTRACT
This
sons aged
65 y surveyed
study
S Haines,
and
Ruth
compares
dietary
1977-78
Nationwide and pork,
Food Consumption Surveys. whole milk, and white bread decreased
in low-fat
beef, pork,
grain
breads. (fruits,
high-fiber
with
and fish, low-fat
consumption
and fat (high-fat
ofcalories fiber
poultry
However,
desserts,
cereals,
milk,
of many butter,
and
increases
and whole-
important and
sources
margarine)
vegetables)
changed
and little
be-
tween 1977 and 1987. The food-consumption trends translated into modest changes in overall nutrient intake. Gender differences were small and contradict the prevailing feeling that women are changing their diets more rapidly than are men. The authors suggest that public health messages have focused too heavily on foods to avoid while not giving adequate guidance for how to plan and prepare meals that will enable older Americans to meet the current diet and health recommendations. Am J C/in Nutr
l992;55:823-30.
KEY derly
WORDS Food consumption people, fat and fiber intake
trends,
nutrition
of el-
popular
press,
government
publications,
and
scholarly
lit-
erature. Nutritional priorities for addressing diet-disease associations have included the reduction ofdietary cholesterol, total fat, and animal fat and an increase in consumption of fruits, vegetables,
whole
grains,
and
evidence
exists
that
in the direction
in serum aged
that
20-74
experts
some and
aged
for
the
advocate.
period
199
in dietary
fiber intake until
the
1).
mid
y (KC
Virtually
because
between
in women
that
increased
their
diet
1960
and
people aged
aged
Reidy,
BM
Popkin,
no information of a lack ofdata
Nuir
al (5) found
Printed
in USA.
will
12.6%
of Amergroup, of 60 million
be in this
that
70%
of healthy
women
and
age
men
cated that they were concerned Patterson and Block (6) studied
about cholesterol how well the diets
in the second
and
vey
National
(NHANES
Health
II) study
Nutrition
(1 976-1980)
>
65 y mdi-
in their
diet.
of participants
Examination
consumed
Sur-
foods
recom-
mended as possibly protective against cancer and reported that older people were closer to the guidelines than younger people and that women were closer than men. However, focus group discussions with older people conducted by the Public Health Service and the Administration on Aging reveal that many people knew
what
not
to eat but
trends
were
unable
to describe
what
consti-
(7).
research
on adult
in an important
segment
women
by examining
ofsociety,
men and women aged 65 y. Because it has that women are leading the way in consuming diets, we examine differences in food-group trends women and men. Finally, we report selected trends to provide a context in which to interpret trends.
Survey
American
suggested more healthful between older
been
nutrient-intake
the food-group
design
sample to in
unpublished
American
1
Society
2
48 coterminous
From
the Department Hill.
Supported
Foundation,
states.
were conducted,
of participants
at Chapel
exists on trends on dietary fiber
© 1992
in the
four waves
1980s.
l992;55:823-30.
19.6%
an increase of 29 million to a total population aged > 65 y (3). Several studies indicate that older people are interested in their health and are willing to change their behaviors to improve their health; some believe their willingness to adopt healthful behavior exceeds that of any other age group (4). Goodwin et
holds
19-50
consumption were important
egg consumption
US population:
of the by 2025,
The Nationwide Food Consumption Surveys of 1977-1978 (NFCS77) and 1987-1 988 (NFCS87) select samples from stratified-area probability samples ofioninstitutionalized US house-
in US adults
Until recently, little attention was paid to the nutritional status and nutrition-related needs of older individuals in this country (JT Dwyer, unpublished observations, 1990). This group is the J C/in
segment 65 y and
Reductions
in the elderly
Similarly,
indication reduced
I 9-50
observations,
Am
are
Methods
fiber.
altered
and fat intake
were greatest women.
of dietary
have
of shifting patterns of food intake from 1977 During the same period, fat intake also decreased
women
in foods
sources
health
reported
declines
y, particularly
components 1985 (2).
other
Americans
concentrations
y were
y there has been oflower-fat foods
adult
public
cholesterol
1980 (1). These 65-74
growing
icans
food-group
Concern for reducing fat and cholesterol intake and improving the healthfulness of diet was a theme of the l980s, reflected by the emphasis on these issues by all forms of media, including
Some
fastest
tuted a balanced diet We build on previous
Introduction
the
98713
E Patterson
practices of perand 1987-88 Intakes ofhigh-fat beef
as part ofthe
1 977-i
by
The
Washington,
For
(8, 9). In NFCS77
ofNutrition, International
NFCS77
one per season,
The University Life
and
NFCS87,
each on a different and NFCS87 the in-
Science
ofNorth
Carolina
Institute-Nutrition
DC.
3 Address reprint requests to BM Popkin, CB# 8120 University Square, University of North Carolina at Chapel Hill, Chapel Hill, NC 27514. Received August 5, 1991. Accepted for publication October 30, 1991.
for Clinical
Nutrition
823
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
Barry
changes
824
POPKIN
El
AL
89.6%
resent
dividual
food-intake
family
members’
vidual
dietary
through odology
phase
covered
intakes intake
were
data
to that
Both
obtained
year, and individual for 3-d periods. Indi-
for 3 consecutive
a mix of 24-h recall of individual dietary
is similar
the entire
days
and 2-d food record. intake measurement
surveys
The methin NFCS87
stratified-area response
(PSUs)
differed
factors
were
designed
samples
of the
rates
for the
as self-weighting,
household
and of
every
sample
level.
proportional
individual
ratio
rate
factors within
it and
aged > 19 y. There in NFCS in 1987 (< 35%)
consider
was than
a much in 1977
comparability
of the
samples
studied
in 1977 and
90.1%
1987,
and
89.8%
of dietary
respectively.
analysis
uses the USDA Nutrient Data Bank, 1987-88, for use with the NFCS87 to compute nutrient of all the dietary data collected in 1977 and 1987. A
specifically
content Linkage
70%). We feel that to consider such data as a large sample of the US population, let alone as representative of the noninstitutionalized older American population, it is important to use the weights for descriptive statistics. (For multivariate analyses where persons with large weights in cells with few cases can inappropriately affect the results, it would be appropriate to exelude the weights.) This is particularly true for trends analyses
where
This
USDA-developed
of the
of fat and
analyses
each
to the colwere applied
in a PSU
individuals
nonresponse
Nutrient
units
The weighting
to the
expected number of completed schedules lected number ofschedules. These weighting undersampling
For
population
significant
to be
multi-
US population.
various
at a statistically calculated
fiber consumed
updated
were
88.3%
linking
(USDA
program
Nutrient
Database
System) was used to match food codes in the 1977 NFCS data with the updated 1987 food codes and nutrient data. The nutrient database update included two types of changes: improvements
in the
quality
of the
food-composition
data
and
real changes 1987 nutrient
in the nutrient composition of foods. Use of the database makes it possible to study trends in food components, such as saturated fatty acids and dietary fiber, that were not available in the original 1977 nutrient database. However, this important advantage is counterbalanced with the bias introduced by concurrent changes in the food supply. Between 1977
and
1987
a frequently
cited
change
in the
food
supply
is
the closer trimming of fat from cuts of meat at the retail level. Use of 1987 nutrient level for 1977 meats will tend to result in a conservative estimate of fat consumption from these sources in 1977.
is important.
Results Sample Food-group We selected
a sample
ofpersons
aged
NFCS77 and NFCS87 obtained l-d food records for up to 3 consecutive all 3 d ofintake
it is recognized
records
that
were
this
may
used
result
analysis.
a l-d, 24-h recall and two days. Only individuals
Both
with
65 y for this
in this analysis. in an
Although
unknown
selectivity
this sample is desirable for two reasons. First, to interpret the percentage ofthe population consuming from a food group, bias,
it is necessary
surveys.
that
a consistent
Second,
variation
is a serious
1 1), and
the
average males and
for
and 618
Food The
in the partment
this
can
unit
shown with
days ofdietary-intake ofpeople
females
(total
females
(total
that
assessing
when
be dealt
analysis
1655
grouping
have
difficulty
problem
of multiple
available
temporal
researchers
aged
2667)
and
be used
for both
intraindividual usual intake (10,
partly
by
use of the data. The 3-d sample 65 y is for 1977, 1012
for 1987,
430
males
1048).
scheme
groups. Initial major used by the USDA.
food groups were Mixed meat-based
based on entrees
were assigned to the appropriate meat category (eg, chicken a la king to chicken) whereas non-meat-based mixed foods were assigned to the grain-based food group (eg, macaroni and cheese). Fat and dietary fiber thresholds were used to separate major food groups into more distinct, nutrient-based food groups. Fifty-six food groups were developed that reflect the total diet. Food group names and characteristic foods appear in Appendix A. The characteristic foods were those most frequently consumed in each
food
in this
analysis.
group. This
A subgroup subset
make substantial contributions food groups in which major
of 43 food
reflects
interest
groups
is presented
in food
groups
that
to fat and dietary fiber intake or trends occurred. These groups rep-
trends
column
gives
at least column by those the 3 d.
one item from the food group over a 3-d period. The for grams per user per day gives average consumption consuming at least one item from a food group over The last two columns show changes or trends in the
percentage
the
proportion
consuming
of elderly
and
gram
people
per
user
who
consumed
1 977
between
and
1987.
Often
it is helpful
population
to have
consumption
10 food-group for adults
food grouping system used separates virtually all foods Individual Food Consumption Surveys of the US Dcof Agriculture (USDA) into 56 useful descriptive and
nutrient-based food groupings
consumption
Overview. The focus of this research is on trends in eating patterns in older Americans from 1977-1978 to 1987-1988. Table I shows the nutrient-based food groups of interest and assesses consumptions in two ways. The percentage-consuming
contributors
aged
a summary
patterns.
65 in 1987
to energy, and
measure
In Table the
to examine
2 we present
total
the top
fat, and dietary
relative
rankings
in
fiber 1977
and 1987. Such a population-based measure reflects not only the contribution offoods that are rich sources ofa nutrient (such as high dietary fiber ready-to-eat cereals) but also the contribution offoods consumed either frequently or in large quantities (such as the contribution of white bread to fiber intake). As expected, the primary sources of energy, total fat, and dietary fiber frequently do not overlap. With minor exceptions, the top 10 sources ofenergy, fat, and fiber in 1977 remain in the top 10 in 1987. High-fat desserts and medium-fat beefand pork remain major sources of calories from 1977 to 1987. However, high-fat beef and pork and eggs fell out ofthe top 10 contributors to calories in 1987. With only minor shifts in rank, butter and margarine, high-fat desserts, and medium-fat beefand pork are the top three contributors to the population fat intake in 1977 and 1987. Sources of dietary fiber remained virtually unchanged from 1977 to 1987, with noncitrus fruits, high-fiber ready-to-eat cereals, whole-grain breads,
and
to fiber
intake.
low-fiber
vegetables
being
the
top
four
contributors
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
stage, survey,
larger (nearly
obtained
of NFCS77.
USDA
to every
were
and
DIETARY TABLE I Food-group
consumption
patterns:
1977-1978
NFCS
CHANGES
and
IN OLDER
1987-1988
NFCS
(weighted)
adults
Amount consumed
Percent consuming
group
Percent consuming
g.uer’.d’
Low-fat milk and milk products High-fat milk and milk products Low-fat cheeses High-fat cheeses Low-fat beefand pork Medium-fat beefand pork High-fat beefand pork Low-fat poultry High-fat poultry Low-fat fish High-fat fish Bacon Low-fat luncheon meats and hot dogs High-fat luncheon meats and hot dogs Butter and margarine Low-fat salad dressings High-fat salad dressings Low-fat desserts High-fat desserts High-fat salty snacks Eggs and egg dishes
30.47 60.81 15.87 34.93 29.06 60.73 38.82 19.02 34.01 15.08 16.91 30.20 13.80 36.88 75.45 2.32 47.13 26.43 62.68 37.23 62.99
216.32 213.98
Legumes
20.04
Low-fat, low-fiber breads Low-fat, high-fiber breads High-fat, high-fiber breads Pasta,rice,andcereals Low-fiber ready-to-eat cereals High-fiber ready-to-eat cereals Low-fat grain mixed dishes High-fat grain mixed dishes
70.46 44.25 8.62 43.84 28.06 28.30 5.18 12.20
79.34 ± 2.81 42.84 ± 0.68 43.57 ± 0.96 40.67 ± 2.71 113.90±2.92 14.86 ± 0.37 25.65 ± 0.83 88.51 ± 7.73 68.46 ± 2.73 142.71 ± 2.46 139.49 ± 2.99 109.60 ± 4.88 179.22 ± 9.30 74.80 ± 1.39 50.52 ± 1.75 61.23 ± 1.53 96.85 ± 1.72 56.63 ± 1.34 469.02 ± 6.61 305.55 ± 6.64 190.68 ± 6.90 205.84±25.92
fruitsandjuices
54.04
Other fruits Noncitrus fruitjuices Fruit drinks Low-fat potatoes High-fat potatoes Green and orange vegetables Low-fiber other vegetables High-fiber other vegetables Coffee
64.68 13.99 12.22 60.32 27.48 43.15 84.79 49.52
Tea
44.87
Regular soft drinks Dietsoftdrinks
16.02 4.54
S
SE. SEs were based
85.87
on weighted
means
Change
6.l3 ± 4.93
Percent consuming
g#{149}user’#{149}d
% ±
Amount consumed
44.31 46.33
245.23 177.57
± 10.26 ± 8.17
%
g#{149}user’#{149}d
13.84 -14.48
56.22
± 2.77
18.71
49.55
± 3.52
2.84
21.55 96.32 69.91 53.55 56.29 56.74 48.47 53.24 12.44 26.98 30.01 10.43 12.37 10.80 81.45 70.72 10.89
± 0.60
35.85
18.63
± 0.64
0.92
3.27
34.16
74.55
± 4.13
5. 10
± 1.29 ± 1.19 ± 1.95 ± 1.26 ± 2.02 ± 1.99 ± 0.45 ± 1.18 ± 0.80
56.90 24.49
59.08 51.60
± 1.71
33.40
55.61
± 2.33
34.65 25.26 14.34 24.49 16.81 35.28
51.96 50.07 50.53 10.30 26.82 27.27
± ± ± ± ± ±
± 0.25
75.13
10.76
± 0.50
± 1.34 ± 0.33
3.22 46.73
12.10 12.84
±
45.04
± 0.73
but not adjusted
±
±
2.64
26.09
1.51 ± 0.37
for design
Proportion consuming. Major shifts in the proportion of the population consuming from food groups are summarized in Table 3. The largest increases were in the lower-fat products and the largest decreases in the higher-fat products. An examination ofthe major shifts in percentages consuming food sources of fat and cholesterol over 3 d reveal interesting changes in behaviors. The largest trends occurred in milk and meat groups. The proportion of US elderly people consuming whole milk decreased whereas the number of low-fat milk con-
53.07 21.1
2.13 2.50 2.81 0.51 1.55 1.05 1.23
± 0.55 ± 3.56 69.63 ± 2.30 13.64 ± 0.66 39.19 ± 1.10 82.64 ± 4.58 41.21 ± 1.24 42.36 ± 1.30 30.06 ± 3.51 129.10±4.20 17.97 ± 1.60 29.01 ± 1.02 94.83 ± 6.41 79.54 ± 4.29 147.05 ± 3.86 138.51 ± 4.04 134.86 ± 9.69 152.61 ± 8.70 72.83 ± 1.97 75.46
61.94 40.58
±
± 3.07
1
54.96 53.41 1 1.92 41.34 23.36 29.05 1 1.33 20.67 52.46 74.61 13.82 14.42 63.37
Amount consumed
-3.83 -14.34 14.38
0.64 10.17 -2.57 -5.72 3.00 -1.60 -0.32 0.91
-0.40 -0.34 -0.74 3.34 -9.92 1.07 -15.50 9.16 3.31 -2.50 -4.70 0.75 6.15 8.47 -1.58 9.93 -0.16 2.20
29.09 -36.41 -6.67
-2.92 -21.77
-10.82 -1.95 -0.68
-4.78 1.60 -2.70 -2.14 -0.16 -2.74 0.34 -0.27
2.04 -5.99 -1.09 2.75 -5.85 3.29 -1.63 -1.21 -10.61 15.20 3.12 3.36 6.32 1 1.09 4.34 -0.99 25.26 -26.60
3.05
-1.97
27.81
45.64
± 1.88
0.33
-4.88
45.04
56.62
± 2.05
1.89
-4.61
88.08 49.83 79.19 39.35 25.51 15.04
95.54 ± 2.67 55.03 ± 2.08 485.25 ± 12.00 297.29 ± 1 1.55 229.26 ± I 1.48 206.79± 13.48
3.28 0.3 1 -6.67 -5.52 9.49 10.51
-1.31 -1.60 16.23 -8.26 38.58 0.95
effects.
sumers increased. High-fat beef and pork consumers decreased whereas there was an increase in the proportion consuming lowfat poultry, low-fat fish, and low-fat beef and pork. There was a decrease in the number ofusers ofthe traditional breakfast items of eggs and bacon. However, little change was observed in the proportion consuming butter and margarine, salad dressing, cheeses, salty snacks, or rich desserts. The trends in eating behavior with regard to fiber intake are less straightforward. The proportion of US elderly people con-
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
%
Citrus
65 y 1987
1977
Food
825
AMERICANS
826
El
POPKIN
TABLE 2 Major sources
ofenergy,
fat and fiber for US men and women
aged
AL
65 y in 1987 with
1977 and
Energy
1987 rankings
Fiber
Fat
Rank Foodgroup
Rank
Rank
87
NFCS87
(weighted)
77
NFCS87
77
NFCS87
%
87
77
%
High-fatdesserts Medium fat beefand pork High-fat, low-fiber breads Noncitrus fruits Low-fat, low-fiber breads (white) Low-fat, high-fiber breads (wholegrain)
8.1 5.3 4.5
1 2 3
1 2 5
4.3
4
7
1 1.1
1
1
4.0
5
3
3.6
10
9
3.8
9.1
4
2
3.5
6 7
9
Butter and margarine
8
9.5
1
2
8 9
4
4.3
7
5 5.7
7
7
6 10
4.9 4. 1
4 8
4 6
4.4
6
7
4.0 4.7
9 5
8 9 10 9.2 9.3 7.7 6.5 4.9 4.8
3 2 5 6 8 9
3 4 5 6 8 10
High-fat milk and milk products (whole milk) Low-fat milk and milk products Low-fat potatoes High-fat beefand pork Eggs and egg dishes Luncheon meats and hot dogs High-fat poultry Salad dressing High-fat cheeses Low-fiber vegetables High-fiber ready-to-eat cereals
High-fiber
3.3 3.1 3.0
9.2 8.2 3.4
2 3 10
3 1
10
other vegetables
Legumes Green and orange vegetables Pasta, rice, and cereals
suming comitant
low-fat, low-fiber bread (white) decreased with a conincrease in low-fat, high-fiber bread (whole wheat). A
greater
number
of elderly
people
also
consumed
higher-fat
low-fiber sweet-type breads. Despite the media to high-fiber ready-to-eat cereals, a very slight was
seen
to-eat
in the
cereals
consuming was
number combined
and
sumption based
with
low-fiber
observed
rice,
of people
in the
cereals.
proportion
consuming
category.
were observed
for foods
category.
groups containing fruit category.
The
dietary
high-fiber
decline
ready-to-eat
cooked-cereal
mixed-dish
consuming
a modest
A very
number
modest
decline
in the
increases
in the low- and high-fat largest
increase
fiber was observed
women.
in users
very little
change.
consumption
citrus
fruit.
Trends
observed in consumption or margarine, or high-fat
10 sources Similarly,
etable
three
product
In general, were
not
shifts large,
in mean
grams
consumed
ofa
population
is but
one
as much
changes. from 1977
aspect
patterns poultry
is the abvirtually of high-fat for men or
accounted
for 3
at 9.2%, 9.5%, and 4.0% of dietary fiber showed in legumes, fruit, and vegexcept
for increases
in non-
per user per day consuming
of consumption
to
with
consumed.
1987.
However,
a few
changes
echoed
the
food-
group consumption findings. For instance, in foods contributing to fat intake, for people consuming high-fat milk, average consumption decreased by 36.4 g. Low- and medium-fat beef and pork consumption decreased by 21.8 and 10.8 g respectively. However, consumption increased by 29.1 g by those using lowfat milks.
in the noncitrus-
categories
of dietary fat many sources
The proportion group
grainof food
in conjunction
of foods
categories per user
Gender
In 1987 these
of the top respectively.
pasta, in con-
Perhaps as important as the trends in consumption sence of change in many product areas. For example, no change was desserts, butter
ready-
in the
foods
However,
but
attention given overall increase
trends in the quantities of specific Changes in mean gram consumed per day are summarized in Table 4 and did not show variability as the results on food-group consumption That is, usual portion sizes did not appear to vary much
considered
food trends.
from This
a food must
be
results
In many areas ofconsumption, men and women had common consumption trends. For example, men and women were likely to report similar trends in beefand pork consumption between 1977
creased
and
1987.
Significant
numbers
of men
and
women
de-
consumption ofthe high-fat meat cuts, with modest increases in consumption ofthe low-fat cuts. Approximately 14% more of both men and women reported consuming low-fat poultry products and neither group reported any decline in consumption ofthe high-fat poultry items. Similar numbers of men and women decreased consumption of egg and egg dishes and bacon products. Men were equally as likely as women to consume fewer white-bread products and more whole-grain-bread products.
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
%
87
DIETARY
CHANGES
TABLE 3 Changes in proportion of US elderly people consuming groups, 1977-1978 to 1987-1988 (weighted)
selected
IN OLDER
food
827
AMERICANS
TABLE
4
Changes
for US elderly
people,
1977-1987
to 1987-1988
Change Change
and food group consuming
> 10%
consuming
Low-fat, High-fat High-fat
in percentage
g. user’ Increases > 10 g.user #{149}d Regular soft drinks Low-fat milk and milk products Noncitrus fruit juices Coffee
14.4 13.8
10.5 10.2 consuming
High-fat
grain-based mixed dishes 10 g.user #{149}d High-fat milk and milk products Fruit drinks Low-fat beef and pork Medium-fat beef and pork High-fat, high-fiber breads
9.5
Decreases
9.2 8.5 6.7 6.2 5.1 consuming 6.7 5.7 5.5
of population
and
14.5
might daily based
15.5 14.5 14.3
and
a number
of gender
trends.
Table
differences
5 identifies
were
the
observed
largest
five servings
products.
in
women. was greater
genders
high-fat
oftotal
Similarly, for men luncheon
fat. Although
meats
declined
meats
numbers
in both
luncheon-meat bers of female more
the decrease between 1977
women
than
are
genders,
consumption consumers. men
one
of the
the
top
conboth
10 sources
ofhigh-fat
luncheon
increase
in low-fat
entire
was attributed to increased numA less healthy trend is that in 1987
shifted
36.4 26.6 21.8 10.8 10.6
to consumption
however,
are considerably
of individuals offruits
and
below
consuming vegetables
declining
intakes
that
the recommended and
in both
servings
men
of grain-
and
women,
of men exceeded the daily threshold of 300 and 1987. The mean cholesterol intake for was < 300 mg in 1977 and continued to de-
in 1987.
differences
in whole-milk and 1987. For
ofconsumers
dine
g for men
Although
cholesterol intakes mg in both 1977
between men and women. In 1977, more women than men consumed low-fat and skim milk, but by 1987 the proportion of men aged 65 consuming low-fat and skim-milk products exceeded sumers
15.7
be expected
women, However,
>
9.9
low-fiber breads milk and milk products beef and pork
consumption
38.6 29.1 25.3 16.2 15.2 1 1.1
Pasta, rice, and cereals
9.9
d’
of high-fat
salty
Discussion The discussion of diet in the aging population often centers on the extent of undernutrition and dietary deficiency and the determinants ofinsufficient food intake (JT Dwyer, unpublished observations, Americans
with
1990). Fewer authors have examined diets of older respect to their healthfulness in light of practices
that may lead to chronic disease. Murphy et al (13) examined diets of individuals aged 55 surveyed as part of the National Health and Nutrition Examination Survey 1971-74 (NHANES I) and the NHANES I Epi-
snack foods. Men continued to consume a higher proportion of dietary fiber from legumes than did women but the differences narrowed between 1977 and 1987. In contrast, the modest increase in
demiologic
numbers
ofconsumers
offsetting
trends
Top gender differences in proportion of US elderly people consuming selectedfood groups, 1977-1978 to 1987-1988 (weighted)
this
product
ofhigh-fiber
in which category
likely
to consume
sume
the
more in
pasta
high-fiber
ready-to-eat women
1987. than
starchy
but
cereals
fewer
reflected
men
consumed
Fewer
men
than
women
but
more
were
likely
rice
vegetables
in 1987
than
sistently
nutrient
healthful
dietary
changes and
guidance
in the
(Table
6). The
sistent with recently published et al (12), but the magnitude 1987
is modest.
saturated 1987,
Although
fat declined values
respectively.
1987
fiber
unweighted of difference and
recommended intakes
suggested absolute
were
by
national
ofenergy
are
con-
by Wright 1977 and from
fat and
women
between 1977 and levels of 30% and 10%,
of 1 1.6 and
12.2
g for women
10 y
Cha nges in percent consuming group
Males
Females %
are con-
values
estimates between
administered
in 1977.
in this sample
direction
the percentage
for men
still exceed Dietary
observed
(NHEFS),
5
Food
mean
Survey
to con-
Nutrients The
TABLE
Follow-Up
Low-fat milk and milk products High-fat milk and milk products Low-fat luncheon meats and hot dogs Low-fat fish High-fat salty snack High-fiber ready-to-eat cereals Pasta, rice, and cereal High-fiber other vegetables
17.8 -
1 8.8
-0.9 14.0 -0.0 -4.1 -7.4 4.3
1 1.0 -
I 1.4
5.8 7.5 5.8 4.2 1.0 -2.6
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
10% increase in percentage of population Low-fat poultry Low-fat milk and milk products Diet soft drinks Low-fat fish 5-10% increase in percentage of population Other fruits Regular soft drinks Low-fat, high-fiber breads High-fat, grain-based mixed dishes High-fat, low-fiber breads Low-fat, grain-based mixed dishes Low-fat beef and pork 5- 10% decrease in percentage of population Eggs and egg dishes Coffee Bacon Tea decrease
Amount
Percent
>
(weighted)
828
El
POPKIN
TABLE Changes
6 in nutrient
intake
for US elderly
people
by gender,
1977-1978
AL
to 1987-1988
(weighted)
Female Nutrient intake
1977
Male 1987
Change
1977
Total 1987
Change
1977
1987
Change
Energy
Dietary
fiber
5876.0 ± 45.6* 1404.4 ± 10.9 58.3 ± 0.5 37.1 ± 0.2 13.5±0.1 46.2 ± 0.2
(g)
Cholesterol
270.5
later.
They
concluded and
were
reported
and
vegetables.
ofsweets
that
cheeses
was little change
± 3.6
the
in overall
were
with two
the
poultry
products,
be asked ofUS
the
if
elderly
to other American population dietary trends ofwomen aged trends levels
in whole and for proportion
comparable It appears into
(2). that
low-fat using
older
fat beef, pork, and agenda for healthful translate reason
into modest that nutrient
contributors bution parent
beef
to fat intake substitution
without
The
fiber
13.6
± 0.2
0.8
266.7
± 4.7
-53.1
intakes
or nu-
but were not adjusted
number
for design
of servings
time
but
of
that
there fruit
of servings
findings proven Except
sizes
in overall
regarding given
et in-
vegetables
behaviors and
fiber
intake
its suggested
health
lack with 1985,
high-
nutritionist’s trends intakes. One the three top
desserts, contri-
cereals,
most
trends
to affect
change
since
to cancer and heavy consuming
for fiber-rich
1977)
and
heart
media more
emhigh-
foods
were
unimpressive.
The gender between
the
conventional trend
setters
results sexes wisdom
were more than
for that
in nutritional
the women change.
remarkable differences (at least
for the similarities and
empirically
1988
USDA
Bridging
Study
was
designed
to
contradict
in this age range)
the are
probes),
food
coding,
weight-to-volume
conversion factors at the food level, and nutrient databases (14). One of the primary research questions of the bridging survey was to determine whether interviewing and coding differences across the surveys resulted in differences in the frequency with which meat and poultry items were reported as eaten without fat or skin. Between the NFCS77 and the NFCS87, coding rules regarding
reported
meat
and
poultry
consumption
were
not
changed. trimmed as “both
That is, responses that did not specify whether fat was from meat or whether poultry was skinned were coded fat and lean eaten” in both surveys. Although the NFCS77 food instruction booklet used by interviewers contained explicit probes for whether the skin on poultry and the fat on meat had been eaten, greater emphasis was placed on these probes in NFCS87. The study suggests that differences in probing contributed to an increase in reporting of0.06 meat and poultry items with fat and/or skin not eaten per person per day (from 0.15 using 1977 interviewing techniques to 0.21 using 1987 interviewing techniques).
items
intake.
(little
The
(including
messages
in their
trends.
consumption
procedures
from
high-fat
related
real
the NFCS77
fat-
at all fat are very
not be adequate
relation
including
to a number
consti-
are similar
away
change
may
of factors,
could be at-
43
compared 1977 and
trends
dietary-fat
1987-1988
test for the effects ofprocedural differences between and NFCS87, specifically differences in interviewing
(or
1987
beef and pork sizes of users
value for constipation, for a handful of women
ready-to-eat
Murphy
we identify groupings.
not
and
changes, changes in survey methodology, and potential nonrandom response bias in the 1987 data. In the examination of trends in dietary fat and food sources offal, it has been suggested that assumptions regarding the coding of meats and poultry may influence the estimated fats and fat-
margarine,
did
in food
1977-1978
tributed
are on any food-consumption
and
pork)
ofdifferences
and
are translating
(butter and
The magnitude
Increases
and
The
observed
trient intakes between
legumes,
food-frequency
milks and and serving
effects.
products.
groups. When 19-50 y between
in portion
changes
surprising
disease, phasis.
± 0.1
± 3.4
from 1977 to 1987 (almost 30%). The apof high-fat items with lower-fat alternatives
alterations
the desired
12.8
319.8
changes in overall nutrient trends may be modest is that
to fat intake
medium-fat
-1.6 -1.1 2.0
1.2
1977
milk consumption change. These
-3.0
-68.9
dietary
behaviors.
± 0.9
36.0 ± 0.2 12.5±0.1 47.3 ± 0.3
± 8.3
all fruits
Americans
food-consumption
64.2
± 0.3
of fish,
between
± 0.6
15.7
of meat
positive
37.6 ± 0.1 13.6±0.1 45.3 ± 0.2
-45.6 -10.9
320.2
In contrast, food
people
67.2
-2.0 -1.4 2.1
± 72.0 ± 17.2
± 0.2
used a food-grouping scheme trends. All milk products, all
and
-3.1
6616.2 1581.3
± 6.1
are difficult.
meat
± 1.5
36.2 ± 0.3 12.4±0.1 46.1 ± 0.4
± 45.2 ± 10.8
14.5
nonquantified
tute 3 of the 18 food groups. sensitive and dietary-fiber-sensitive
76.3
6661.8 1592.2
389.1
study
present
dissimilar,
± 1.0
63.6 15.2
0.6
in number
in the two surveys and many important dietary
It may
38.3 ± 0.2 13.8±0.1 44.0 ± 0.3
± 120.5 ± 28.8
-42.1
reported
struments that hides
thereof)
79.7
-1.3 -0.9 1.9
alcohol.
Comparisons
and
-2.8
7828.7 1871.1
± 4.9
over
servings
± 78.7 ± 18.8
± 0.2
means
servings
Declines
and
± 0.9
7765.1 1855.9
12.2
slightly
capita
-126.4 -30.2
228.4
per capita
increased
in per
al (1 3) used
are
55.5
SE. SEs were based on weighted
all milks
7.0 ± 16.7 ±
35.8 ± 0.3 12.6±0.1 48.1 ± 0.3
1 1.6 ± 0.1
(mg)
S
and
5749.6 1374.2
Similarly,
a decline
introduced
as eaten
with
was
observed
in meat
and
poultry
skin not specified (from 0.42 per day with 1977 methods to 0.32 with 1987 methods). These results suggest to us that differences in fat probing may have contributed to some overestimation of fat intake from meat consumption in 1977. However, from these results it is also clear that it would be inappropriate to assume that all or even most of meat and poultry consumed in 1977-1978 was either skinned or trimmed, as was suggested (CE Woteki, MG Kovar, H Riddick, unpublished observations, 1984). In addition, we can examine the bridging survey for evidence of bias
reported
because
fat and/or
of differences
in nutrient
databases
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
(U) (kcal) Fat (g) Fat (%) Saturatedfattyacid(%) Carbohydrate (%)
DIETARY
CHANGES
IN OLDER
and the weight-to-volume conversions. Ifwe examine the extent of difference in overall fat intake attributed to use of 1987 vs 1977 nutrient composition and weighting factors, the estimate for women aged 20-49 y is 24.1 vs 24.3 g fat at the 10th percentile, 59.2 vs 63. 1 g at the 50th percentile, and 123.2 vs I 21.2 g at the
90th
percentile.
We
estimate
that
this
translated
to a
ofhouseholds aged 15-24
aged < 41 y and without y. Although no exact figures
children, and participants are available for response
rates ofelderly people, it appears reasonable to assume that they are fairly well represented except for the very poor or very rich. As for self-selection bias, if respondents and nonrespondents have systematically different behaviors, then survey results may be influenced. Bethlehem (16) showed how reweighting can reduce the potential for nonresponse bias, and all results reported here are weighted. It remains possible, however, that respondents with
demographic
and
socioeconomic
characteristics
identical
to those of nonrespondents have systematically different eating behavior. Although caution must be exercised when interpreting these data, as noted by the US General Accounting Office (17), the NFCS87 is the only survey available on individual diets over the last decade and the data are critical for many users. In conclusion, results of focus groups of older Americans mentioned
in the
introduction
are
reflected
in results
of this
study. Elderly Americans knew what not to eat but were not able to describe what constitutes a healthy diet. In this study a number ofpositive dietary changes are observed, particularly in foods where the presence of fat is obvious (eg, meats, poultry, and dairy products). However, this risk-avoidance behavioral pattern is not accompanied by strong, consistent, health-seeking behaviors, ie, consumption ofmore fruits, vegetables, and whole grains. Perhaps the diet and health messages have focused too heavily on foods to avoid while not giving adequate guidance for how to plan, shop for, and prepare healthful meals that will enable individuals to meet the current diet and health recom-
These
results
are consonant
with
Improving America our attention on the need
Medicine
which
focuses
hesive
and
practical
approach
to
a recent
Institute
‘s Diet and Health, to develop a more co-
report
nutrition
education
for
the
LI
country(l8).
We acknowledge the assistance ofSuzanne Harris and the International Life Science Institute-Nutrition Foundation Subcommittee on Nutrition and Aging in developing this research. We also thank them for assistance in clarifying some ofthe important issues. We thank David Guilkey for his assistance in addressing issues of weights and Phil Bardsley for his extensive computer assistance. References 1 . National
Center for Health Statistics. Trends in serum cholesterol levels among US adults aged 20 to 74 years. JAMA 1987;257:93742. 2. Popkin BM, Haines PS, Reidy KC. Food consumption ofUS women: patterns
and
determinants
between
1977
and
1985.
Am
J Clin
Nutr
l989;49: 1307-19. 3. Martin LG. Population aging policies in East Asia and the United States. Science l991;l51:527-31. 4. Heckler MM. Health promotion for older Americans. Public Health
Rep l985;I00:225-30. 5. Goodwin JS, Leonard AG, Hooper EM, Garry PJ. Concern about cholesterol and its association with diet in a group ofhealthy elderly. Nutr Res 1985;5:141-8. 6. Patterson BH, Block G. Food choices and the cancer guidelines. 7.
Am J Public Health l988;78:282-6. 5K. Aging and health promotion:
market research for public
Maloney
education. Executive Service, 1984.
summary.
Washington,
DC:
Public
Health
Peterkin B, Rizek RL, Tippett KS. Nationwide food consumption survey, 1977. Nutr Today 1988{n/Feb: 18-24. 9. Rizek RL. The 1977-78 nationwide food consumption survey. Fam 8.
Econ Rev l978;4:3. Beaton GH, Milneri, Corey P, et al. Sources ofvariance in 24-hour dietary recall data: implications for nutrition study design and interpretation. Am J Gin Nutr l979;32:2546. 1 1. Hartman AM, Brown CC, Palmgren J, et al. Variability in nutrient and food intakes among older middle-aged men. Am J Epidemiol l990;l32:999-1012. 10.
12.
Wright
HS,
Guthrie
HA,
Wang
MQ,
Bemardo
V. The
1987-88
nationwide food consumption survey: an update on the nutrient intake of respondents. Nutr Today May/June:2l-7. 13. Murphy SP, Everett DF, DresserCM. Dietary patterns. In: CornoniHuntley JC, Huntley RR, Feldman ii, eds. Health status and well. being of the elderly. National Health and Nutrition Examination Survey-I
epidemiologic
follow-up
versity Press, 1990:185-209. 14. Guenther PM, PerloffBP. 1977
and
estimates Bridging
1987
in the
study.
Effects
Nationwide
of food and nutrient Study. Washington: US
(NFCS report 15. Life Sciences
87-M-1.) Research
New
York:
Uni-
ofprocedural differences between Food Consumption Survey on intakes: results of the USDA 1988 Department
ofAgriculture,
Office. Impact of nonresponse Nationwide Food Consumption
data from the 1987-88 thesda, MD: Federation
Oxford
ofAmerican
Societies
1990.
on dietary Survey.
for Experimental
BeBi-
ology, 1991. 16. Bethlehem JG. Reduction of nonresponse bias through regression estimation. J Offic Stat l988;4:25l-60. 1 7. US Government Accounting Office. Nutrition monitoring: mismanagement of nutrition survey has resulted in questionable data. Washington,
DC:
GAO,
18. Institute of Medicine. ington DC: National
1991.
Improving Academy
(GAO
RCED-91-l
America’s Press,
1991.
17.)
diet and health.
Wash-
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
difference in energy from fat of 146.4 Id (35 kcal) at the 50th percentile. However, the estimates of energy intake at the 50th percentile for this cohort of women also differed, from 6259.3 Id (1496 kcal) using 1987 nutrient-composition data to 6322.0 Id (151 1 kcal) using 1977 data, making estimates of percent of energy from fat closer than might appear from just examining the grams of fat differences. We have chosen not to attempt to adjust the values for selected food codes, in part because we do not know the true nature and extent of other food-supply changes and in part because of a desire to maintain methods that might be easily replicated by other investigators. We believe the methodologic advantages of using 1987 nutrient-composition data outweigh disadvantages. The bias introduced in the estimation of fat and fat-containing food trends is that the method we use produces more conservative estimates of fat intakes in 1977 (2% lower), which will bias our hypotheses of declining fat intakes over time towards the null. Another concern in trends analyses is the low response rate for NFCS87. Participation by households drawn into the sample was < 35%, primarily because of heavy respondent burden and low remuneration. Low response rates may affect the accuracy of a survey estimate through a decrease in the sample and potential self-selection bias. The Life Sciences Research Office (15) reviewed the impact of nonresponse and noted low representation from economically poorer and richer households, households with working female heads ofhouseholds or female heads
mendations. of
829
AMERICANS
830
POPKIN
APPENDIX Food
system
Food
group
Low-fat High-fat Low-fat High-fat
High-fat
beefand
Low-fat High-fat
poultry poultry
milk products milk products
skim
foods
milk
milk cheese
American, cheddar Ham, beef roast (lean only) Lean ground beef, beef roast (fat content not specified) Regular ground beef, beef pot roast Skinned chicken breast, turkey Chicken or turkey salad, chicken or turkey pot pie, fried chicken leg (not specified as to skin) Tuna (canned)
pork and pork
fish fish
Food
group
Representative
Legumes Low-fat, Low-fat,
low-fiber high-fiber
breads
High-fat,
low-fiber
breads
breads
Low-fiber
ready-to-eat
cereals
High-fiber
ready-to-eat
cereals
Low-fat, dishes
grain-based
meats
mixed
grain-based mixed dishes Citrus fruits and juices Other fruits High-fat,
meats
dressings
desserts desserts
High-fat salty snacks Eggs and egg dishes
Cornbread,
Oatmeal, Cornflakes, Shredded wheat Spaghetti sauce
and
Pork sausage, bologna, beef hot dog Margarine, butter Low-calorie French and Italian dressings Mayonnaise, Italian and French
Gelatin
dessert
Ice cream, apple pie, butter or sugar cookies, cakes Saltines, potato chips Fried egg, boiled egg, omelet
bran
muffin
rice
wheat flakes wheat, bran cereal, flakes with tomato and meat
Macaroni
and
cheese,
thin-crust
pizza Orange juice
Banana,
apple, strawberries,
cantaloupe
Noncitrus fruit juices Fruit drinks
Apple juice, Cranberry
cranberry cocktail,
apple juice drink, Low-fat
potatoes
High-fat
potatoes
Green
and orange
Low-fiber
other
vegetables
vegetables
dressings
Low-fat High-fat
Pinto beans, pork and beans White bread Wheat or cracked wheat bread
flavored
dressings
salad
and
method not specified) Bacon Sliced ham, turkey ham
foods
Rolls (white, soft)
High-fat, high-fiber breads Pasta, rice, and cereals
Fried perch, fish (cooking
Butter and margarine salad
2% Milk,
Whole Cottage
pork
Bacon Low-fat luncheon hot dogs High-fat luncheon hot dogs
High-fat
Representative
High-fiber Coffee
other vegetables
Tea soft drinks Diet soft drinks Regular
juice cranberry-
fruit-
drink
Baked potato, mashed potato Home fries, potato salad, French fries Carrots, broccoli, spinach Lettuce, tomatoes (raw), lettuce salad, onions, celery (raw), green beans Green peas, corn Coffee (ground and instant) Tea Soft drink (cola) Soft drink (cola, sugar-free)
Downloaded from https://academic.oup.com/ajcn/article-abstract/55/4/823/4694328 by East Carolina University user on 03 January 2019
milk and milk and cheeses cheeses Low-fat beef and Medium-fat beef
Low-fat
AL
A
grouping
Low-fat High-fat
El