RESEARCH ARTICLE

.............................................................................................

Variability in Percent Energy from Fat throughout the Day:

Implications for Application of Total Diet Goals

RACHEL BALLARD-BARBASH, I FRANCES E, THOMPSON, I

BARRY I. GRAUBARD,2 AND SUSAN M. KREBS-SMITHI

Applied Research Branch 1 and the Biometry Branch,2 DiVision of Cancer Prevention and Control,

National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892

ABSTRACT In spite of the general agreement that dietary recommendations apply to the diet over time, the quantitative levels for total and saturated fat have been used m estabhshmg federal policy related to mtakes for a single day, meal, and even an mdivldual food. Application of these recommendations uniformly to meals across a day Implies that fat intake IS uniform throughout the day. ThiS analysis of the 1985 Con­tinumg Survey of Food Intakes by Individuals demonstrates that percent energy intake from fat across eating occasions Within a day is not uniform. Percent energy intake from total and saturated fat IS lower at the morning meal and at snacks among women at all levels of fat consumption, suggestmg that fat is restricted more often at these two eating occasions. Intake of total and saturated fat was also more variable at these two eating occasions. These findings suggest that restricting fat mtake at these eatmg occasions and liberahzing fat intake at midday and evening meals occurs commonly and may be an effective fat-reduction strategy. Daily variability in percent energy from fat should be considered in designmg dietary fat reduction interventions and m applymg quantitative recom­mendations for percent energy from total and saturated fat m nutrition guidance directed to individual meals.

UNE 26:278-283, 1994)

INTRODUCTION

Quantitative goals of 30% or less of total calories from fat and less than 10% of calories from saturated fat have been established by both federal agencies and voluntary health organizations. I-6 These goals are based on research that has examined dietary intakes of foods and nutrients over time and related these intakes to growth, health, and chronic diseases. Statements within the Diet and Health Report and the Dietary GUIdelines for Americans indicate that the quantitative recommendations for total and saturated fat

Address for correspondence Rachel Ballard-Barbash, M D, MPH., Apphed Research Branch, DIVISIOn of Cancer Prevention and Control, National Cancer InstItute, ExecutIve Plaza North, Room 343, 6130 ExecutIve Blvd MSC 7344, Bethesda, MD 20892-7344, Tel (301) 496-8506, Fax (301) 496-9949 © 1994 SOCIETY FOR NUTRITION EDUCATION

278

intake are intended to apply to dietary intake over time. I.2

Accordingly, some national surveys designed to monitor nutrient intake have estimated usual intake from averaging data on intake over several days.7.8

In spite of the general agreement that these recommen­dations apply to the diet over time, the quantitative levels for fat and saturated fat have been used in establishing federal policy related to intakes for a smgle day, meal, and even an individual food. 9

.IO For example, while the 1993 food label­

ing regulations developed by the Food and Drug Admini­stration (FDA) did not directly use the quantitative percent energy from fat goals for the total diet as the criterion for a low total or saturated fat nutrient content claim on individ­ual foods,9 they used the total diet goals as the baSIS for deriving gram cutpoints to define foods carrying a low-fat nutrient content claim. Similarly, the 1992 Nutrition Guid­ance for the Child Nutrition Programs developed jointly by the Departments of Agriculture and Health and Human ServIces stated that the quantitative fat goals "apply to the diet over several days, not to a single meal or food."IO Because of this underlymg premise, suggestions are given for calculating the average percent energy from fat by summing fat intake for lunch menus over several days rather than from a single day. However, the 30% or less of total calories from fat and less than 10% of calories from saturated fat were applied to both breakfast and lunch individually m this guidance. This suggests that it was assumed that percent energy from fat would be relatively constant across meals throughout the day.

Interventions to reduce total fat intake generally recog­nIze that intake varies throughout the day. Often these mter­ventions suggest that people achieve overall lower intake of fat by eating very low-fat breakfasts and snacks, thereby allowing more liberal intake of fat during the larger midday and evenmg meals. 4. 11-13 We were unable to find pubhshed research in free-living populations focusing on fat content of various meals, on whether fat mtake for individuals varies throughout the day, and on whether this variability is associated with the overall level of fat intake. In addition, there are no published descriptive reports on which meals

Journal of NutTltlOn EducatIOn Volume 26 Number 6

tend to be lower in fat. If individuals who are most success­ful in achieving recommended levels of total and saturated fat intake do so by eating much lower levels of fat at some meals, such as breakfast or snacks, this finding mIght have relevance to federal nutntion guidance and programs.

This analysis describes percent energy intake from total and saturated fat for various eating occasions by levels of daily percent energy intake from total and saturated fat Intake averaged over several days of data collection and examines within-person variabilIty in fat intake across eat­Ing occasions throughout the day.

METHODS

Sampling and study procedures. Data from the USDA's Continuing Survey of Food Intakes by IndivIduals 1985 (1985 CSFII) Core Sample were used. The design provided a multistage, stratified area probability sample of women aged 19 through 50 and their children aged 1 through 5 years, representative of the 48 coterminous states.

Survey participants were asked to provide six nonconse­cutive I-day dietary recalls collected at approximately 2-month intervals throughout a year. Interviews occurred from April 1985 through March 1986. Eligible house­holds were scheduled for interviews on different days of the week to provide representativeness of dietary intake data over all days of the week. The first interview was conducted in person; subsequent Interviews were con­ducted by telephone.

While the design of the survey called for collection of 6 nonconsecutive days of information, less than half of the 1985 CSFII respondents interviewed initially completed the full 6 days of report. Consequently, USDA constructed a sample of all women who had completed any 4 of the possible 6 days of report, 71 % of the Initial 1985 CSFII sample. For all selected women, the first day of information was included. For those women who reported 5 or 6 days of information, USDA randomly selected the 3 additional days included. The dataset used in these analyses IS com­posed of 1032 women, each with 4 days of dietary informa­tion. A more complete description of the sampling and study procedures is found in CSFII report 85-4.7 Data from children aged 1 to 5 were not analyzed.

Definition of meals. For each dietary recall, women reported all foods and beverages consumed throughout the day, including the time each was consumed and a name for each separate eating occasion. Names of eating occasions were chosen by the respondent from a list and included breakfast, brunch, lunch, dinner/supper, snack/beverage break (party, happy hour), or something else.

The respondent-provided information about the name and the time of the eating occasion was combined to form analytical meal categories. The data showed that breakfast was commonly used to describe the first eating occasion of

November. December 1994 279

the day and that snacks were uniquely defined and were not commonly considered a "meal." However, terms for mid­day and evening meals, such as brunch, lunch, dinner, and supper, were used interchangeably, dependent on regional and cultural differences. Therefore, the timing of the eating occasion was given precedence over the name of the meal in defining the midday and evening meals. The morning meal category included eating occasions defined as breakfast regardless of time and brunch if before 10 a.m. The midday meal category included eating occasions defined as brunch if at or between 10 a.m. and 3:30 p.m. and lunch, dinner, or supper if before 3:30 p.m. The evening meal category included eating occasions defined as lunch, dinner, or sup­per if at or after 3:30 p.m. The snacks category included eating occasions defined as snack/beverage break or some­thing else, regardless of when it was consumed. This cate­gorization of eating occasions generated 16,512 potential eating occasions: four possible eating occasions for each of 1032 women over 4 days of data collectlOn.

Exclusions. Although the unit of analysis was individual women, some meals and days of data were excluded from an individual's record in order to create a dataset that contained only eating occasions that included caloric con­sumption. Therefore, 4 of 4128 days of data recording were excluded due to no calories being consumed on those days. In addition, days when women reported being sIck or ill were excluded. This exclusion removed 214 of 4124 days and 571 of 13,371 meals. Of the 16,512 potential eating occasions in this dataset, women reported consumption of any food or beverage item on 13,371 eating occasions. For 3141 potential eating occasions, primarily snacks, women reported no calories; only two were occasions in which foods or beverages without calone content were consumed. Be­cause this analysis focuses on variabilIty across eating occa­sions, days In which women reported one eating occaSlOn were excluded from all analysis, removing 76 of 391 0 days and 76 of 12,800 eating occasions. Finally, mne women with only 1 day of data after the above exclusions were excluded from the analysis. The final analytic sample consisted of 1023 women who had 12,699 eating occasions over 3825 days of data collection. The distribution of study days by the number of eating occasions and by day of data collection is shown in Table 1. The distribution of the women by the number of days of data per respondent was 77.7%, 18.5%, and 3.8% with 4, 3, and 2 days of data, respectively.

Statistical analysis. A dataset was constructed that in­cluded, for each woman, the nutrient contributions of all meals for each day of intake. For these analyses, nutrients of interest are energy (kcal), fat (gm), and saturated fat (gm). Percent energy from total and saturated fat were computed from these data for each meal.

For the purpose of creating descriptive data on typical meal Intakes of percent energy from total and saturated fat, mean values for percent energy from total and saturated fat

280 Ballard-Barbash et al.lVARIABILITY IN PERCENT ENERGY FROM FAT

Table 1. Distribution of study days by the number of eating occasions

and day of data collection among 1023 women from the 1985 CSFII.

Day of Data Collection

Number of Eating

Occasions 2 3 4 Total

Two 118 134 110 119 481

Three 416 411 434 378 1639

Four 444 422 404 435 1705

Total 978 967 948 932 3825

for each of the four types of eating occasions were calcu­lated. These mean values were calculated for each person for each of the four possible eating occasions by summing individual eating occasion values for percent energy from total (or saturated) fat and dividing by the number of eating occasions available over the 4 days of data collection.

In order to describe the variability in percent energy fat (or saturated fat) intake across eating occasions within a day, two measures of daily variability in eating occasions were calculated: (1) a dally coefficient of variation (CV) to measure percent variabilIty about the mean, and (2) a daily standard deviation (SO) to measure absolute variabIlity about the mean. In this analysis, these measures are pre­sented as descriptive measures of variability and are not used to test hypotheses about population variability. The CV among eating occasions for a given day was calculated for each person by first computing the mean eating occasion percent energy from total (or saturated) fat by summing the percent energy from total (or saturated) fat from each eating occasion in a day and dividing by the number of eating occasions (up to four) in that day . The CV for a given day was then calculated as the SO of an individual's mean eating occasion percent energy from total (or saturated) fat divided by the individual's mean eating occasion percent energy from total (or saturated) fat for a given day. Finally, a mean daily CV was calculated by summing the daily CVs for each individual and dividing by the number of days contributing to the calculation for that individual. A mean daily SO was calculated for each indiVIdual by summing the dally SOs from an individual's mean eating occasion percent energy from total (or saturated) fat and dividing by the number of days contributing to the calculation for that individual. To adjust for the potential effect of calorie intake on percent energy intake from fat, a parallel analysis was run in which these two measures of variability were computed using weights based on the calorie level of each meal.

Analyses stratified by usual percent energy from fat over the 4 days of records were run to examine whether variabil­ity in fat intake across eating occasions varied by amount of usual fat intake. For these stratified analyses, usual percent

energy from total and saturated fat was calculated by sum­ming each individual's intake of total (or saturated) fat in grams over the available days, multiplying by 9, diVIding by the sum of the kcals over the available days, and multiplying by 100.

Weighting factors that adjust for differential probability of selectIOn and household and indiVIdual nonresponse were applied. The analyses were run uSing SUOAAN, a software package appropriate for analysis of surveys WIth complex sample designs .14

RESULTS

The distribution of the study days by the number of eating occasions and the days of data collection (Table 1) indIcate that three and four eating occasions per day were the most common eating pattern, allowing assessment of variability In percent energy from fat across eating occasions. The most frequent eating pattern (45%, 1705 of 3825 days) was four eating occasions per day . The pattern of three eating occa­sions per day was nearly as common (43%, 1639 of 3825 days). The pattern of two eating occasions per day was reported for 13% (481 of 3825) of the days of the survey. The number of study days included in this analysis declined over the 4 days of the survey. The major exclUSIOn respon­SIble for thIS decline was the increaSing number of women excluded for reporting that their Intake was atypical due to being ill on the day of reporting (35, 45, 61, and 73, respectively, over the 4 days of the survey) . In addition, slightly more women were excluded during the later days of the survey for reporting only one meal per day.

Mean percent energy from both total and saturated fat varied by meal (Table 2). The mornIng meal and snacks were similar In percent energy from total and saturated fat, 27 .6% and 26.3% for total fat and 11.4% and 10.9% for saturated fat from breakfast and snacks, respectIvely, and lower compared to the mIdday and evening meals . The midday and evening meals were similar In percent energy from both total and saturated fat, 37.5% and 38.9%, and 13.0% and 13.6%, respectively. This pattern of differences In percent energy from fat was also present when the cohort was stratified by percent energy intake from total and saturated fat.

The SO of mean percent energy intake from total and saturated fat for each of the four eating occasions also varied, with snacks having the highest SOs for both nutrients (of 16.6% for total fat and 7.5% for saturated fat). Midday and evening meals were somewhat similar and had lower SOs of mean percent energy intake from total and saturated fat . Values for SDs can be derived by multIplying the SEs in Table 2 by the N of each group.

The daily variability among eating occasions in percent energy from fat and saturated fat as measured by the mean daily CV (Table 3) was hIghest among the group with the lowest fat intakes and decreased as percent energy Intake

Journal of Nutntlon Education Volume 26 Number 6 November. December 1994 281

Table 2. Mean Intake of total and saturated fat by meal and by level of percent energy from fat over the 4 days of records.

Eating Occasions

Usual % Energy Morning Midday Evening Snacks

Total fat Mean percent energy from total fat (N)

All 27.6 ± .61 37.5 ± 4 38.9 ± .4 26.3 ± 6

(1003) (1002) (1019) (991)

:;;30 17.2 ±.6 28.7±1 .1 31.3 ±.9 17.6 ± 2.0

(123) (124) (126) (120)

30.1 - 39.9 26.9 ±.6 36.6 ± .4 37.3 ± .4 25.1 ±.7

(563) (560) (572) (562)

~40 33.0 ± 1.1 42.6 ±.6 44.7±.4 32.0 ± 1.3

(317) (318) (321) (309)

Saturated fat Mean percent energy from saturated fat (N)

All 11 .4 ± .3 t3.0 ±.2 136 ± .2 10.9 ± .3

(1003) (1002) (1019) (991)

:;;10 6.9 ± 7 88 ± .3 9.9 ± .3 6.3 ± .7

(123) (124) (126) (120)

10.1 -15.9 11 .0 ±.2 129 ± 2 133 ±.2 10.6 ±.3

(563) (560) (572) (562)

~ 16 15.8 ±.8 16.5 ± .4 17.7 ±.3 15.3 ± 1.1

(317)

1Standard error of the mean = SO divided by N.

from fat and saturated fat increased. Adjusting for calorie content of individual meals, which caused the midday and evening meals to contribute most to the estimation of variability, resulted in lower CVs. However, the trend of declining CVs with increasing percent energy from total and saturated fat remained: 43.9%, 32.1%, and 26.0% for total fat and 47.2%, 36.1 %, and 31.4% for saturated fat for low, middle, and high intakes of fat and saturated fat, respectively .

Conversely, the daily variability among eating occasions in percent energy from fat and saturated fat as measured by the mean daily SO (Table 3) was lowest among the group with the lowest fat intakes and increased as percent energy intake from fat increased. Adjusting for calorie content of individual meals again resulted in lower SOs, but the trend of increasing SOs with increasing percent energy from total and saturated fat remained: 10.5%, 10.8%, and 10.7% for total fat and 3.7%, 4.6%, and 5.3% for saturated fat for low, middle, and high intakes of fat and saturated fat, respectively.

DISCUSSION

Dietary guidance recommendations have applied quantita­tive fat intake goals intended for diet over time to individual

(318) (321) (309)

meals over time. 10 This application of the quantitative goals for percent energy from fat may in part be based on an assumption that percent energy from fat is uniform through­out the day. Data on the potential differential biological effect of marked variatton versus uniformity in fat intake throughout the day are not available but ideally would be considered prior to assuming that percent energy from fat should be umform throughout the day. This analysis dem­onstrates that percent energy from fat is not uniform across meals throughout the day among this representative sample of free-living women. Total and saturated fat intakes were much lower at the morning meal and at snacks. The very low percent energy from fat intakes for these two eating occasions among women with the lowest overall fat and saturated fat intakes, similar to quantitative total diet goals, suggests that restricting fat intake at these meals and liber­alizing fat intake at midday and evening meals is an effective strategy to reduce percent energy from fat. The variability in percent energy from fat intake across meals among women meeting dietary recommendations demonstrated that percent energy intake from fat does not need to be low at every meal in order to achieve the currently recom­mended levels of 30% or less of energy intake from fat . However, given that women with higher fat intakes also demonstrated variability in intake across meals, variable

282 Ballard-Barbash et aJ.lVARIABILITY IN PERCENT ENERGY FROM FAT

Table 3. Mean daily variability' in percentage of energy from total and

saturated fat intake within a day by level of usual percentage of energy

from total and saturated fat.

Daily Variability in % Energy from Fat

Usual % Energy

from Fat (n)

Total fat

All (1023)

:<;30 (126)

30.1 - 39.9 (575)

~40 (322)

Saturated fat

All (1023)

:<;10 (118)

10.1 -15.9 (716)

~16 (189)

Coefficient of

Variation (CI?

56.7

73.0 (67.7, 78.3)

56.2 (54.6, 57.8)

50.9 (48.6, 53.2)

62.8

77.7 (72.6, 82.8)

61.9 (60.4, 63.4)

55.8 (52.6, 59.0)

Standard Deviation

(CI)

16.1

16.0 (14.6,17.4)

16.3 (15.8,16.7)

17.5 (16.7,18.3)

70

5.7 (5.2, 6.2)

6.9 (6.7, 7.1)

8.4 (7.9, 8.9)

'Daily variability is a measure of variability across the four possible

eating occasions within a day as defined for this analysis.

2CI = 95% confidence intervals.

restriction in fat intake is unlikely to be a sufficient behavior to achieve recommended total dietary goals without an overall reduction in fat mtake.

As anticipated, this analysis demonstrated that variability in fat intake across eatmg occasions throughout the day varies over levels of percent energy from fat. When the CV is used to assess percent variability about the mean, variabil­ity increases as percent energy intake from fat decreases. This suggests that mdividuals with lower fat intakes restrict their fat intake to a greater extent at some meals than do individuals with higher fat intakes. The generally lower levels for percent energy from total and saturated fat at the morning meal and at snacks among all individuals suggest that fat is restricted more often or to a greater extent at these two eating occasions. These findings are consistent with dietary fat reduction interventions that recommend that people achieve overall lower intake of fat by eating very low-fat breakfasts and snacks, thereby allowing more liberal intake of fat during the higher calorie midday and evening meals. 'I - IJ However, when SO, a measure of absolute vari­ability about the mean, is used, variability does not increase as percent energy intake from fat decreases but rather decreases slightly. It is not clear which of these measures of variability is the most appropriate to describe variability in fat intake. Because stratification of percent energy from fat resulted in distributions within strata that were not normally distributed, especially in the extreme strata, both the CV and SO were correlated with mean values for percent energy from fat in the extreme strata when examined by

Pearson correlations. For example, correlation coefficients between mean percent energy from fat and CV and SO of mean percent energy from fat were - .39% and .24%, re­spectively, among those with usual percent energy from fat of ~ 30%, and -.58% and .17% among those with usual percent energy from fat of ~ 40%. Because the measure of variability used should ideally be independent of the mean, differences in these measures of variability across levels of fat intake must be interpreted with caution.

Although the mean daily variability in percent energy from fat declined when adjusted for calorie content of eating occasions, the pattern of increasing CVs and decreas­ing SOs with decreasing percent energy intake from fat remained. The Impact of the calorie content of an eating occasion on variability in fat intake is also evident in the higher SOs about mean percent energy from fat observed for morning meals and snacks, meals that were much lower in calorie content than midday and evening meals. (The mean calorie contents of the morning meal and snacks were 292 and 336 kcals, respectively, compared to the mean calorie contents of the midday and evening meals of 523 and 697 kcals , respectively.) The higher SOs about mean percent energy from fat for morning meals and snacks may reflect the greater overall variability, in terms of calones and types of foods consumed, for these meals compared to midday and evening meals.

Snacks are particularly variable in content, ranging from the consumption of a single caloric beverage to numerous food and beverage Items. It is possible that the increased CV in percent energy from fat among the group with lower fat intakes was due to the occurrence of more frequent and smaller eating occasions with greater variability in fat and calorie content in this group compared to those with higher fat mtakes. However, in this sample, the frequency distri­bution of eating occasions did not differ across levels of percent energy from total and saturated fat (data not shown). In fact , the percentage of eating occasions that are snacks among those 10 the groups with the lowest percent energy from total and saturated fat was slightly lower than that of those with higher fat intakes, 22.5% versus 23.5% for total fat and 22.4% versus 23.4% for saturated fat .

In additlon, in this analysis, the term encompassing snacks included all eating occasions not included under the definition of morning, midday, or evening meals. There­fore , the term "snacks" for this analysis includes both single and multiple eating occasions over a day. Forty-one percent of the snacks included only a single eating occasion; 86% included three or fewer eating occasions. In order to exam­ine whether the pattern of increasing CV with decreasing fat intake varied among women with different snacking patterns, the sample was stratified by predominant snacking patterns: zero or one snack per day on all days, two or three snacks per day on any or all days, and four or more snacks per day on at least 1 day. The pattern of increasing CV with decreasing fat intake was found irrespective of whether the

Journal of Nutrition Education Volume 26 Number 6

snack term included single or multiple eating occasions over a day (data not shown).

The dietary patterns of individuals among the group with lower fat intakes may not necessarily reflect current dietary recommendations. Others have examined this issue in this survey data and found that women in the lowest fat intake group consumed more alcohol, were less likely to meet the RDA for zinc and calcium, and were significantly more likely to meet the RDA for ascorbic acid than women in the groups with higher fat intake. 15 The analysis presented here did not include, for example, an examination of whether alcohol or added caloric sweetener consumption was greater for persons with lower fat mtakes. Alcohol and added caloric sweeteners can decrease the percentage of calories from fat without actually improving the overall diet quality, as they increase the total and the nonfat calories but add few nutrients. Although the calories from these two components are included in the calculations of total energy in this survey, estimates of their individual contributions to intake are not generally calculated because alcohol intakes are assumed to be somewhat under-reported and there are not complete data on the proportion of carbohydrates in each food that come from added sweeteners.

It is possible that demographic and other nondietary factors might influence total and saturated fat intake l5 ,16 and variability in fat intake. In this analysis, no associations were found between age, body mass index, physical activity, and self-reported health status and percent energy from total and saturated fat and variability in percent energy from fat. Therefore, it seems unlikely that these factors explain the observed changes in variability in percent energy from fat as percent energy from total and saturated fat increases.

CONCLUSIONS

Dietary interventions, dietary guidance, and the application of dietary guidance to federal food service programs are most likely to be effective in promoting healthier eating patterns when they reflect patterns based on known effec­tive dietary interventions and incorporate existing dietary behaviors where possible. II This analysis suggests that vari­ability in percent energy from total and saturated fat across the day should be considered in designing dietary fat reduc­tion interventions and applying quantitative recommenda­tions for percent energy from total and saturated fat to nutrition guidance directed at individual meals.

ACKNOWLEDGMENT

We wish to acknowledge Lisa Licitra Kahle (Information Management Services) for expert computer assistance.

November. December 1994 283

REFERENCES

1. NatIOnal Academy of Sciences. NatIOnal Research CounCil Diet and

health: ImphcatlOns for reduclllg chromc disease risk Washington,

DC: NatIOnal Academy Press, 1989.

2. United States Departments of Agnculture and Health and Human

ServlCes. Nutrltion and your health: dietary guidelines for Amencans.

3rd ed. Home and Garden Bulletlll No. 232. Washlllgton, DC: U.S.

Department; of Agnculture and Health and Human Services, 1990.

3. Butrum RR, Chfford CK, Lanza E. NCI dietary gUidehnes: ratIOnale.

Am] Chn Nutr 1988; 48:888-95

4 NatIOnal Institutes of Health. NatIOnal Cholesterol EducatIOn Pro­

gram Report of the Expert Panel on PopulatIOn Strategies for Blood

Cholesterol ReductIOn. NIH PubhcatlOn No. 90-3046. Washmgton,

DC: U.S. Department of Health and Human SerVices, 1990.

5. Amencan Heart ASSOCIatIOn. Amencan Heart ASSOCIation dietary

gUidehnes for healthy Amencan adults Dallas, TX: Amencan Heart

ASSOCIatIOn, 1988

6 Amencan Cancer Society. Amencan Cancer SOCiety nutntlOn and

cancer: cause and preventIOn Amencan Cancer Society SpeCial Re­

port. Atlanta, GA: Amencan Cancer SOCiety, 1984.

7. Umted States Department of Agnculture. Human NutntlOn Informa­

tIOn SerVlCe. NatIOnWide food consumptIOn survey' contlllulllg survey

of food llltakes by llldlviduals, women 19-50 years and their children

1-5 years, ... days, 1985. CSFII Report Nos. 885-4. HyattSVille, MD:

US Department of Agnculture, 1987.

8. Umted States Department of Agnculture. Human NutntlOn Informa­

tion Service. NatIOnWide food comumptlon survey: contlllulllg survey

of food mtakes by llldividuals, women 19-50 years and their children

1-5 years, 4 days, 1986. CSFII Report Nos. 86-3. HyattSVille, MD:

US Department of Agnculture, 1988.

9. Department of Health and Human Services. Food and Drug Admllli­

stratlOn Food labehng proposals. Federal Register 21 CFR Part 101,

et aI, Part III. Washmgton, DC. U.S. Department of Health and

Human ServlCes, 1991

10. Umted States Department of Agriculture Food and Nutntlon Service.

BUlldmg for the future. nutntlOn gUidance for the Child NutntlOn

Programs. US Department of Agnculture, FNS-279. Alexandna, VA:

U.S. Department of Agnculture, 1992.

11. Food and NutntlOn Board. Institute of Medlcllle, NatIOnal Research

CounCil. Improvlllg America's diet and health: from recommendatIOns

to actIOn. Washlllgton, DC: National Academy Press, 1991.

12. National Cancer Institute. Fat budgetlllg sectIOns of the Polyp Pre­

ventIOn Trial. NutntlOmsts' and PartiCipants' Manuals for the Polyp

PreventIOn Tnal, National Cancer Institute, 1991.

13. Goor R, Goor M, Boyd K. Eater's chOice: a food lover's gUide to

lower cholesterol. Boston, New York: Houghton Mlffim, 1987.

14. Shah BV. SUDAAN: profeSSIOnal software for survey data analysis for

multi-stage sample deSigns. Research Tnangle Park, NC: Research

Tnangle Institute, 1993.

15. Sowers MF, Thompson FE, Frongillo EA, Parpla B. Charactenstlcs of

Amencan women aged 19 through 50 years consumlllg a lower fat

diet. Ecol Food and Nutr 1992; 29:1-9.

16. Thompson FE, Sowers MF, Frongillo EA, Parpla B]. Sources of fiber

and fat in diets of women aged 19-50: ImplicatiOns for nutntlOn

educatIOn and pohcy. Am] Pub Health 1992; 92:695-702.