Foods commonly eaten in the United States, 1989-1991 and 1994-1996: Are portion sizes changing?☆

Methods 

Sample 

Estimates of portion sizes were based on intakes by individuals aged 2 years and older who provided all requested days of dietary intake information for the CSFII 1989-1991 and CSFII 1994-1996. For CSFII 1989-1991, intakes were collected on three consecutive days, with a 24-hour recall on the first day followed by day 2 and day 3 records. For CSFII 1994-1996, two nonconsecutive 24-hour recalls were collected 3 to 10 days apart, not on the same day of the week. The surveys have been described in detail previously [29], [30].

Both surveys were designed to provide estimates of food intake by individuals in a nationally representative sample of households in the United States. Breastfed children were excluded from the analysis, and the resulting sample sizes were 11,488 for 1989-1991 and 14,262 for 1994-1996. However, the sample sizes vary for each food in the results tables (Tables 1 and 2) because they are based on the number of individuals who consumed each food and not on the total sample size for a particular age-sex group.

Table 1. Foods with significant differences in amounts consumed per eating occasion, 1989-1991 and 1994-1996: Differences in amounts and 5th and 95th percentile amountsa

	Food or food typeb	Amount per eating occasion (g)		Difference			Percentiles
							1989-1991		1994-1996
		1989-1991	1994-1996
		←mean (SEM)c,e→		Gramsc,d	%	Common household units	5th	95th	5th	95th
	Whole grain and ″wheat” bread	46 (1)	50 (1)	+4.2	9.1	1/6 slice	24	72	24	92
	Crackers	22 (1)	26 (1)	+3.2	14.3	1 saltine cracker	6	56	6	62
	Cookies	36 (1)	40 (1)	+4.2	11.6	2/5 medium cookie	10	84	9	96
	Ready-to-eat cereal	49 (1)	54 (1)	+4.7	9.6	1/12 cup raisin bran or 1/6 cup crispy rice	17	111	18	113
	Corn flakes	40 (1)	46 (1)	+6.3	15.9	1/4 cup	17	74	17	100
	Toasted oat rings	37 (1)	42 (1)	+5.0	13.7	1/6 cup	13	74	14	83
	Rice	165 (3)	150 (3)	−15.3	9.3	1/10 cup	36	390	27	334
	Pasta	140 (3)	162 (3)	+17.9	12.4	1/8 cup spaghetti	30	319	26	420
	Macaroni and cheese	295 (11)	244 (9)	−50.8	17.2	1/5 cup	81	647	53	556
	Spaghetti with tomato sauce	376 (10)	436 (16)	+59.5	15.8	1/4 cup	123	741	122	983
	Pizza	197 (7)	169 (5)	−27.8	14.1	1/3 slice (12-in thin crust)	63	497	36	422
	Lettuce	36 (1)	41 (1)	+4.7	13.0	3/5 leaf	8	93	7	110
	Raw carrots	42 (2)	33 (1)	−8.4	20.4	1/15 cup, chopped	7	120	5	100
	French fries	77 (1)	83 (1)	+5.9	7.7	1 medium french fry	28	117	28	140
	Orange juice	241 (4)	268 (4)	+27.6	11.5	1 fl oz	124	490	124	498
	Raw bananas	101 (1)	111 (1)	+9.4	9.3	1/12 medium banana	49	129	55	136
	Bacon	22 (1)	20 (0.3)	−2.2	9.8	2/5 thin strip	8	47	5	42
	Chicken	107 (2)	95 (1)	−12.1	11.3	2/5 oz	36	234	28	204
	Coffee	380 (6)	463 (10)	+83.3	21.9	2.8 oz	180	718	178	1,066
	Coffee, with caffeine	391 (7)	469 (10)	+81.3	21.0	2.7 oz	179	733	178	1,130
	Tea	351 (7)	397 (10)	+44.1	12.5	1.5 oz	120	719	118	930
	Soft drinks, all	363 (4)	423 (6)	+59.6	16.4	2.0 oz	182	582	184	861
	Soft drinks, sweetened, with caffeine	376 (6)	442 (7)	+66.1	17.6	2.1 oz	186	599	185	926
	Soft drinks, sweetened, caffeine-free	345 (6)	389 (6)	+43.3	12.5	1.4 oz	124	602	151	739
	Soft drinks, diet, with caffeine	363 (7)	420 (8)	+57.0	15.7	1.9 oz	180	553	179	870
	Soft drinks, diet, caffeine-free	329 (8)	368 (8)	+39.3	11.9	1.3 oz	139	480	133	707
	Fruit drinks	319 (6)	373 (8)	+51.0	15.8	1.6 oz	125	596	124	925
	Beer	655 (25)	885 (32)	+230.3	35.2	7.7 oz	225	1,718	315	2,371
	Wine	197 (13)	239 (8)	+42.8	21.8	1.5 oz	5	457	59	518
	Margarine, all types, separately	12 (0.2)	10 (0.2)	−2.3	18.8	1/2 tsp	2	29	1	29
	Margarine, stick	12 (0.2)	10 (0.3)	−2.1	17.8	1/2 tsp	2	28	1	28
	Margarine, soft	12 (0.3)	10 (0.4)	−2.2	18.4	1/2 tsp	2	28	1	29
	Mayonnaise	13 (0.2)	12 (0.3)	−1.3	9.9	1/3 tsp	3	28	2	28

aData shown for all subjects aged 2 years and older. Data presented represent only the foods that had a significant (P=.001) change over time. bFor descriptions of foods included, see Table notes in references 3 and 4. cSample size varies because data are based on the number of individuals consuming each food. d(+) indicates larger amount in 1994-1996, and (−) indicates a smaller amount. eSEM=standard error of the mean.

Table 2. Foods with significant differences in amount consumed per eating occasion, 1989-1991 and 1994-1996, by age and sexa

	Food or food typeb	←Significant difference (g)c→
		All ages	2-5 y	6-11 y	12-19 y		20-39 y		40-59 y		60+ y
		M & F	M & F	M & F	M	F	M	F	M	F	M	F
	Whole grain and ″wheat” bread	+4.2
	(nbsp;(1 slice=26 g)
	Crackers	+3.2
	(4 saltine crackers=12 g)
	Cookies	+4.2
	(1 medium cookie=10 g)
	Ready-to-eat cereal	+4.7		+6.1	+12.7				+10.6	+7.4
	(1 cup raisin bran=56 g)
	Corn flakes	+6.3			+18.3
	(1 cup=25 g)
	Toasted oat rings	+5.0		+11.3
	(1 cup=30 g)
	Rice	−15.3
	(1/2 cup cooked=79 g)
	Pasta	+17.9
	(1 cup cooked spaghetti=140 g)
	Macaroni and cheese	−50.8					−117.0
	(1 cup=243 g)
	Spaghetti with tomato sauce	+59.5			+177.0
	(1 cup=248 g)
	Pizza	−27.8		−31.1								−56.3
	(1 slice (1/8) of a 12-inch=71 g)
	Lettuce	+4.7								+8.2
	(1 cup=55 g)
	Raw carrots	−8.44
	(1/2 cup chopped=73 g)
	French fries	+5.9	−9.9				+15.0
	(10 medium=50 g)
	Orange juice	+27.6							+48.0
	(6 fl oz=187 g)
	Raw bananas	+9.4					+12.0			+11.6
	(1 medium=118 g)
	Bacon	−2.2
	(1 thin strip, cooked=5 g)
	Chicken	−12.1	−16.7	−14.6			−25.9
	(1 oz cooked=28 g)
	Coffee	+83.3					+103.2	+90.9	+123.4	+76.2	+62.0	+46.9
	(8 fl oz=237 g)
	Coffee, with caffeine	+81.3					+100.8	+83.8	+118.6	+65.9	+66.9
	(8 fl oz=237 g)
	Tea	+44.1						+56.2
	(8 fl oz=237 g)
	Soft drinks, all	+59.6				+51.1	+100.9	+70.8	+70.2	+35.9
	(1 12-fl oz can=355-369 g)
	Soft drinks, sweetened, with caffeine	+66.1				+56.0	+94.8	+65.2	+81.3
	(1 12-fl oz can=369 g)
	Soft drinks, sweetened, caffeine-free	+43.3				+63.2		+65.4
	(1 12-fl oz can=369 g)
	Soft drinks, diet with caffeine	+57.0						+84.4	+77.8	+49.7
	(1 12-fl oz can=355 g)
	Soft drinks, diet, caffeine-free	+39.3								+58.8
	(1 12-fl oz can=355 g)
	Fruit drinks	+51.0			+118.7			+89.0
	(8 fl oz=248 g)
	Beer	+230.3							+243.1		+241.7
	(1 12-fl oz can=360 g)
	Margarine, all types, separately	−2.3		−2.9				−2.4	−3.4
	(1 tsp=5 g)
	Margarine, stick	−2.1										−2.3
	(1 tsp=5 g)
	Margarine, soft	−2.2
	(1 tsp=5 g)
	Mayonnaise	−1.3
	(1 tablespoon=14 g)

aSignificance level set at P=.001. bFor descriptions of foods included in the food types, see Table notes in references 3 and 4. cSample size varies within age/sex group because data are based on the number of individuals consuming each food. M=male. F=female.

Results 

Significant differences in portion sizes (ie, amounts consumed per eating occasion) were reported in 1989-1991 and 1994-1996 for about one third of the 107 foods/food types that were examined. Foods with significantly different portion sizes are shown in both Table 1 and Table 2. Table 1 presents data on portion sizes consumed by the total population 2 years of age and older. Table 2 presents data by age and sex groups to see whether differences in amounts consumed were consistent across the population.

Data in Table 1 are presented as mean amounts consumed per eating occasion in 1989-1991 and 1994-1996, as differences between the two surveys, and as 5th and 95th percentile values in both surveys. Smaller portion sizes were reported in 1994-1996 for a few foods, including two mixed dishes (macaroni and cheese, pizza) and bacon, chicken, margarine, and mayonnaise. For the majority of foods with significant differences between surveys, the portion sizes were larger in 1994-1996. Seven of these foods were grains and cereals, and 11 were beverages. Five of the 11 beverage groups were types of soft drinks: all soft drinks, diet with and without caffeine, nondiet with and without caffeine. Coffee, tea, fruit drinks, beer, and wine were also consumed in larger amounts in 1994-1996. The 5th and 95th percentile values for ready-to-eat cereals were remarkably similar in 1989-1991 and 1994-1996, and percent difference was relatively small. The 95th percentile values for beverages were generally higher in 1994-1996, with the overall percent differences of 40% to 60% for most beverages, 80% for coffee, and over 200% for beer.

Table 2 presents data on portion sizes with significant differences by age and sex group. For some of the foods (eg, whole grain and ″wheat” bread, crackers, mayonnaise), portion sizes were different for the total population (Table 1) but not for any specific age or sex group. No foods showed differences for all of the age and sex groups in this study. Differences for some food types were reported for only one or two groups of people. For example, portion sizes were larger in 1994-1996 for cornflakes but only for males aged 12 to 19 years; portion sizes for beer were larger for males 40 to 59 years and 60 years and older. However, for several foods, portion sizes consumed by several age and sex groups were significantly different in 1994-1996. Coffee amounts were greater for all adults aged 20 years and older; soft drink portion sizes were larger for females aged 12 to 59 years and for men aged 20 to 59 years. Generally, the significant differences for a given food were in the same direction for all age and sex groups. The exception is for french fries, which showed an increase for the total population and for males 20 to 39 years old and a decrease for children 2 to 5 years old.

Discussion 

The purpose of this report was to compare the quantities of portion sizes of commonly eaten foods reported in CSFII 1989-1991 and 1994-1996. Several public health and food marketing trends provide a strong argument for examining trends in portion sizes over time. A major public health concern is the increasing prevalence of overweight in the United States [31], [32], [33]. Several investigators have shown that food portion sizes are positively related to energy intake in children and adults [13], [34], [35], [36]. Increased energy intake over recent decades is consistent with trends in overweight ((27)). It is also consistent with trends in marketing and availability of foods. Recently, Young and Nestle reported that marketplace food portions are consistently larger than in the past ((37)).

Despite the relatively short time span between the two surveys, there were differences in amounts of some foods consumed by selected age and sex groups. The differences could affect dietary quality. Smaller portion sizes of margarines and mayonnaise could have contributed to small decreases in the percent of energy provided by total fat in 1994-1996 ((26)). Larger portion sizes of selected ready-to-eat cereals could contribute to intakes of many nutrients [38], [39]. In their analysis of CSFII 1989-1991 data, Subar and colleagues ((38)) showed that primarily because of fortification, ready-to-eat cereals were, for adults, among the top 10 food sources for 18 of 27 nutrients examined.

Portion sizes of soft drinks were larger in 1994-1996 for a number of sex and age groups. Considerable attention has focused on soft drink consumption and consequent implications for nutrient intakes [40], [41]. Most studies of soft-drink consumption and dietary adequacy have examined intake by children and adolescents. Harnack and colleagues ((42)) reported that children's and adolescents' energy intake was positively associated with nondiet soft drinks. Other investigators have reported trade-offs between soft drink intake and milk intake with consequent effects on calcium intake [43], [44].

Larger portions of other beverages may also have dietary and health effects. Larger portion sizes of beer were consumed in 1994-1996 by men aged 40 years and older. The relationship between alcohol intake and energy intake and body weight is complex. However, higher beer consumption was associated with a larger waist-to-hip ratio in both white and black men and women ((46)). Energy derived from alcohol contributes to an increased total energy intake with no compensatory decrease in energy from macronutrient sources ((47)).

Larger portions of caffeine-containing beverages by older adults may also have health effects. Recently, Rapuri and colleagues ((48)) reported a higher rate of bone loss at the spine in postmenopausal women with caffeine intakes >300 mg/day, equivalent to 514 g (ie, a little more than 2 cups) of brewed coffee. Susceptibility to loss was accelerated in women with a specific vitamin D receptor gene variant. At any rate, the amounts cited in their article are similar to mean amounts reported in 1994-1996 and well below the 95th percentile values. In an editorial accompanying the article by Rapuri and colleagues, Massey ((49)) indicated that prudent recommendations to older persons should emphasize moderate caffeine consumption as well as adequate calcium intake.

Given the differences in portion sizes between 1989-1991 and 1994-1996, it is important to consider whether any differences might be the result of methodological differences between the two surveys. As noted in the description of the methods, the 1989-1991 data are drawn from three consecutive days of intake and the 1994-1996 data are for two nonconsecutive days. Although the difference in number of days could certainly impact the number of users, it should not have much effect on the portion of food eaten at a meal. Also differences in the age distribution of users could contribute to difference in portions, particularly if very young children were to make up a larger proportion of the users in one survey than in the other. However, the use of weighted data tends to counter this effect.

Reporting amounts of foods and beverages is difficult and subject to inaccuracies [50], [51], [52], [53]. In general, the surveys were similar in training interviewers and assisting respondents in providing food portion estimates. Interviewers in both surveys were trained to use a food instruction booklet to aid the respondent in reporting amounts of foods eaten [29], [30] The booklets specified the appropriate types of measures for the foods. In both surveys, respondents received a set of measuring guides, including measuring cups, spoons and a ruler, to help them estimate amounts of foods and beverages consumed. They could also use their own household cups, bowls, and glasses.

There was one difference in portion aids that could have affected reporting accuracy for chicken portion sizes between the surveys. In 1994-1996, a visual aid was added to display both a full chicken breast and a half breast, and a whole chicken leg and the drumstick and thigh components. Other possible factors for differences include changes in the food coding database, or more specifically, the measure-to-gram-weights conversions. For example, the standard weight for a medium banana changed from 114 to 118 g between the survey years. Although this discrepancy may have contributed to a small increase in the portion size of bananas, it is unlikely that it would explain the rather substantial increase observed in this study. Additionally, the weights of standard food types such as a medium apple or banana in the database may vary from the apples and bananas in the marketplace, but this would have been an issue for both survey years.

The food coding databases for each survey reflect the marketplace at that point in time. Another type of change between 1989-1991 and 1994-1996 could account for some of the change in the portion size of macaroni and cheese. In the 1980s, much of the macaroni and cheese consumed was homemade. In the 1990s, macaroni and cheese made from dry mix was more popular. This consumption shift could account for a part of the portion size change because the cup weight of the prepared dry mix is 26 to 52 g lighter than that of the homemade product.

Any change in the combination of foods included in a food type across surveys could affect estimates of portion size. The ready-to-eat cereal type includes cereals such as corn flakes, which weigh about 25 g per cup, and raisin bran, which weighs about 56 g per cup. We examined whether the larger portion sizes of ready-to-eat cereals consumed in 1994-1996 could represent proportionately greater use of heavier cereals and thus may have contributed to the higher gram weights of the mean portions. However, on examination of the frequency of reporting of cereals, the patterns of usage are very similar in the two surveys for the most frequently reported cereals, leading us to believe that the increase in portions is not due to consumption of heavier cereals.

There are limitations to using the data in this article to interpret the impact of overall differences across surveys on nutrient intakes and health status. The analyses are based on only those foods identified as most commonly eaten. The extensive variety of foods eaten by Americans limits the number of foods that can be analyzed to ensure statistical reliability. Nevertheless, these data identified changes in food portion sizes that occurred in a relatively short time period. Furthermore, larger portions in the absence of fewer foods or fewer eating occasions may have profound effects on the increasing prevalence of overweight and obesity. Future data on consumption patterns by Americans should be monitored to determine any changes that have occurred since the 1994-1996 study. Changes in portion sizes consumed should be monitored to assess the need for revisions in policy and diet assessment protocols.

Applications 

■Portion size data available from national surveys are widely used for formation of public policy for counseling and dietary guidance of individuals. The findings in this study indicate that portion sizes have changed in recent years for many commonly eaten foods. Some of the changes can adversely affect nutritional well-being and public health of Americans. Some of the changes reflect a marketplace that promotes larger portions. The challenge to dietitians and other healthcare providers is to counsel on the importance of decreasing portions and recognition of appropriate portion sizes.