You are on page 1of 6

ARTICLE IN PRESS

JOURNAL OF
FOOD COMPOSITION
AND ANALYSIS
Journal of Food Composition and Analysis 20 (2007) 733738
www.elsevier.com/locate/jfca

Short Communication

Comparison of proximate, mineral and vitamin composition of common


Brazilian and US foods
Renata M. Padovani, Dag M. Lima, Fernando A.B. Colugnati, Delia B. Rodriguez-Amaya
Nucleo de Estudos e Pesquisas em Alimentac- ao, Universidade Estadual de Campinas, 13083-852 Campinas, SP, Brazil
Received 10 January 2006; received in revised form 16 March 2007; accepted 29 March 2007

Abstract
Because food analysis is costly and laborious, analytical data are frequently borrowed from tables of developed countries and
incorporated in tables of developing countries. Taking advantage of the new Brazilian table of food composition, which is based on
representative national sampling and actual analyses, an attempt is made to gain an insight into the adequacy of this practice by
comparing data from the Brazilian and US tables in terms of the proximate, mineral and vitamin composition of 19 selected foods,
common to both countries. For proximate composition, data agreement is excellent for dry whole milk; good for condensed milk, dry
nonfat instant milk and canned peas; and fairly good for black beans, milk chocolate, lentils, oats, long-grain white rice and wheat our.
Greater variation is seen with minerals and vitamins. In terms of minerals, agreement is very good only for dry whole milk, good for
black beans and canned peas; fairly good with milk chocolate, lentils, wheat our and natural yogurt. For vitamins, agreement can be
considered good only for wheat crackers and yogurt and fairly good for condensed milk and dry whole milk. The comparison indicates
that data may be borrowed for proximate composition provided the foods are equivalent. However, caution must be taken in borrowing
data for micronutrients.
r 2007 Elsevier Inc. All rights reserved.
Keywords: Proximate composition; Mineral content; Vitamin content; Brazilian foods; US foods

1. Introduction
Food analysis is expensive, often complicated and
requires substantial human and material resources. Because of nancial constraints, data are often borrowed
from tables of developed countries, particularly from the
United States Department of Agricultures food composition tables, and incorporated in tables of developing
countries. Such a practice, however, can be questioned
considering that foods vary naturally in their composition.
Numerous papers have shown that the composition of
foods of plant origin can vary signicantly because of
varietal differences, maturity at harvest, climatic effects,
Corresponding author. Departamento de Ciencia de Alimentos,
Faculdade de Engenharia de Alimentos, Universidade Estadual de
Campinas, C.P.6121, 13083-862 Campinas, SP, Brazil.
Tel.: +55 19 35214013; fax: +55 19 35212153.
E-mail address: delia@fea.unicamp.br (D.B. Rodriguez-Amaya).

0889-1575/$ - see front matter r 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jfca.2007.03.006

composition of the soil, production techniques, etc. On the


other hand, the composition of foods derived from animals
is inuenced by the animals breed and age, rearing
technique, composition of the feed, etc. For processed or
prepared foods, aside from compositional variation of the
raw materials, differences in composition can arise from
differences in the amounts of ingredients used, and
especially for labile compounds, the processing or cooking
conditions and the duration and conditions of storage.
Thus, analytical data obtained with food samples of one
country may not be relevant to another country.
The Brazilian TACO (Tabela de Composic- ao de
Alimentos) project recently launched a food composition
table, based on representative national sampling and
analyses carried out by laboratories approved in interlaboratory evaluations (NEPA/UNICAMP/MS, 2004).
The aim of the present paper is to compare the proximate,
mineral and vitamin compositions, taken from the Brazilian and US food composition tables, of 19 selected food

ARTICLE IN PRESS
R.M. Padovani et al. / Journal of Food Composition and Analysis 20 (2007) 733738

734

Table 1
Comparison of proximate composition of common Brazilian and US foods per 100 g of edible portion
NDB no.a

Description

Country

Water
(g)

Energy
(kJ)

Protein
(g)

Fat
(g)

Carbohydrate(g)

Dietary
ber (g)

Ash
(g)

16014

Beans, black, mature seeds, raw

18137

Cake, white, dry mix, regular

19120

Candies, milk chocolate

08020
08506
18157

Cereals ready to eat, corn akes


Cookies, chocolate wafers

11172

Corn, sweet, yellow, canned, whole kernel, drained solids

18232

Crackers, wheat, regular

16069

Lentils, raw

20499

Macaroni, dry, unenriched

01095

Milk, canned, condensed, sweetened

01092

Milk, dry, nonfat, instant, with added vitamin A

01090

Milk, dry, whole

20038

Oats

11308

Peas, green, canned, regular pack, drained solids

20444

Rice, white, long-grain, regular, raw, unenriched

11887

Tomato products, canned, paste, with salt added

11888

Tomato products, canned, puree, with salt added

20481

Wheat our, white, all-purpose, unenriched

01116

Yogurt, plain, whole milk, 8 g protein per 8 oz

Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDAb
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA

14.88
11.02
1.00
4.00
1.26
1.50
4.27
3.53
1.17
4.50
76.22
76.92
4.06
3.10
11.47
10.40
10.24
9.90
26.96
27.16
3.10
3.96
2.70
2.47
9.13
8.22
80.14
81.70
12.90
11.62
79.66
73.50
90.79
87.88
12.98
11.92
90.04
87.90

1383
1427
1760
1782
2287
2238
1640
1502
2102
1812
418
339
1806
1979
1451
1477
1553
1553
1308
1343
1513
1498
2078
2075
1648
1628
315
289
1501
1527
255
343
117
159
1508
1523
215
255

21.34
21.60
6.16
4.50
5.48
7.65
4.33
6.25
5.56
6.60
3.23
2.62
10.06
8.60
23.15
25.80
10.00
13.04
7.67
7.91
34.69
35.10
25.42
26.32
13.92
16.89
4.60
4.42
7.47
7.13
2.43
4.32
1.36
1.65
9.79
10.33
4.06
3.47

1.24
1.42
6.13
10.90
30.27
29.66
0.67
0.75
24.67
14.20
2.35
1.00
14.44
20.60
0.77
1.06
1.30
1.51
6.74
8.70
0.93
0.72
26.90
26.71
8.50
6.90
0.38
0.35
0.27
0.66
0.19
0.47
Tr
0.21
1.37
0.98
3.04
3.25

60.46
62.36
85.21
78.00
61.32
59.40
89.26
87.56
67.54
72.40
17.14
18.59
68.73
64.90
63.86
60.08
77.94
74.67
57.00
54.40
53.04
52.19
39.18
38.42
66.64
66.27
13.81
12.58
78.81
79.95
14.96
18.91
6.89
8.98
75.09
76.31
1.92
4.66

21.8
15.2
1.7
0.9
2.2
3.4
2.1
2.6
1.8
3.4
4.6
2.0
2.5
4.5
16.9
30.5
2.9
3.7
0.0
0.0
0.0
0.0
0.0
0.0
9.1
10.6
5.1
4.1
1.8
1.3
2.8
4.5
1.0
1.9
2.3
2.7
0.0
0.0

3.79
3.60
2.00
2.70
1.68
1.78
1.48
1.90
1.06
2.00
1.06
0.7
2.72
2.80
2.61
2.67
0.51
0.88
1.63
1.83
8.23
8.03
5.80
6.08
1.81
1.72
1.44
0.97
0.54
0.64
2.75
2.80
0.95
1.28
0.77
0.47
0.94
0.72

Data taken from the Brazilian (NEPA/UNICAMP/MS, 2004) and USDA (2005) food composition tables. Values in bold represent appreciable
differences.
a
USDA Nutrient Data Bank number.
b
The values represent the means of NDB no. 08020 and 08506.

products that are common to both countries and appeared


to be equivalent from the description of the products. It is
hoped that the comparison will give an insight on the
validity of data borrowing.
2. Materials and methods
Through the descriptions of the products in the Brazilian
(NEPA/UNICAMP/MS, 2004) and the US (USDA, 2005)
food composition tables, the following food products
commonly consumed in both countries were considered
to be equivalent and were therefore chosen for the
comparison: black beans, white cake dry mix, milk
chocolate candy, corn akes, chocolate wafers, canned
sweet corn, regular wheat cracker, lentils, macaroni,

sweetened condensed milk, fortied dry nonfat milk, dry


whole milk, oats, peas, long-grain white rice, tomato paste,
tomato puree, wheat our, plain yogurt. These products
were compared in terms of the proximate, mineral and
vitamin compositions.
The production of the USDA food composition
database has been widely described (Pehrsson et al., 2000,
2003). For the products compared, the Brazilian sampling
plan covered nine cities in the ve ofcial Brazilian
geopolitical regions (North, Northeast, South, Southeast
and West Central) (Galeazzi et al., 2002). The samples
consisted of the leading brands (maximum of ve brands
for each food), collected from supermarkets/hypermarkets,
where 8485% of total food purchases are made in Brazil.
Two units from different factory lots of each brand of each

ARTICLE IN PRESS
R.M. Padovani et al. / Journal of Food Composition and Analysis 20 (2007) 733738

735

Table 2
Comparison of mineral composition of common Brazilian and US foods per 100 g of edible portion
NDB no.a

Description

Country

Ca
(mg)

Mg
(mg)

P
(mg)

Fe
(mg)

Na
(mg)

K
(mg)

Cu
(mg)

Zn
(mg)

16014

Beans, black, mature seeds, raw

18137

Cake, white, dry mix, regular

19120

Candies, milk chocolate

08020
08506
18157

Cereals ready to eat, corn akes


Cookies, chocolate wafers

11172

Corn, sweet, yellow, canned, whole kernel, drained solids

18232

Crackers, wheat, regular

16069

Lentils, raw

20499

Macaroni, dry, unenriched

01095

Milk, canned, condensed, sweetened

01092

Milk, dry, nonfat, instant, with added vitamin A

01090

Milk, dry, whole

20038

Oats

11308

Peas, green, canned, regular pack, drained solids

20444

Rice, white, long-grain, regular, raw, unenriched

11887

Tomato products, canned, paste, with salt added

11888

Tomato products, canned, puree, with salt added

20481

Wheat our, white, all-purpose, unenriched

01116

Yogurt, plain, whole milk, 8 g protein per 8 oz

Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDAc
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA

111
123
59
192
191
189
56
3
23
31
2
5
20
49
54
56
17
18
246
284
1363
1231
890
912
48
54
22
20
4
28
29
36
13
18
18
15
143
121

188
171
28
11
57
63
8
8
48
53
20
20
40
62
94
122
28
53
22
26
109
117
77
85
119
177
23
17
31
25
29
42
15
23
31
22
11
12

471
352
333
337
212
208
43
35
124
132
61
65
148
220
368
451
100
150
187
253
1673
985
1242
776
153
523
79
67
107
115
47
83
30
40
115
108
119
95

6.46
5.02
1.21
1.39
1.58
2.35
3.90
30.57
2.44
4.01
0.59
0.86
2.20
4.40
7.05
7.54
0.88
1.30
0.13
0.19
0.93
0.31
0.52
0.47
4.45
4.72
1.39
0.95
0.82
0.80
2.09
2.98
1.25
1.78
0.95
1.17
Tr
0.05

o0.4
5
463
664
77
79
405
716
137
580
260
323
854
795
o0.4
6
7
5
94
127
432
549
323
371
5
2
372
252
1
5
498
790
104
399
1
2
52
46

1416
1483
ndb
117
355
372
52
82
240
210
162
195
181
183
887
955
147
162
329
371
1556
1705
1132
1330
336
429
147
173
61
115
680
1014
308
439
151
107
71
155

0.827
0.841
0.153
0.081
0.313
0.491
0.040
0.0915
0.263
0.463
0.046
0.058
0.180
0.318
0.833
0.519
0.150
0.283
0.033
0.015
0.157
0.041
0.110
0.080
0.440
0.626
0.138
0.082
0.147
0.220
0.203
0.365
0.093
0.287
0.150
0.144
0.018
0.009

2.85
3.65
0.58
0.46
1.06
2.01
8.48
0.19
0.87
1.09
0.50
0.39
1.14
1.60
3.49
4.78
0.78
1.26
0.86
0.94
3.84
4.41
2.73
3.34
2.63
3.97
0.88
0.71
1.19
1.09
0.37
0.63
0.33
0.36
0.83
0.70
0.44
0.59

Data taken from the Brazilian (NEPA/UNICAMP/MS, 2004) and USDA (2005) food composition tables. Values in bold represent appreciable
differences.
a
USDA Nutrient Data Bank number.
b
ndnot determined.
c
The values represent the means of NDB no. 08020 and 08506.

product were taken at each sampling site. For each


product, the collected units (totaling 4290 units per
product with total weight of 5410 kg) were sent by air
(under refrigeration for yogurt) to the Instituto de
Tecnologia de Alimentos, Campinas, where they were
mixed, homogenized and packed in lacquered (epoxy)
tinplate cans of 100200 g, and promptly sent to the
laboratories for immediate analyses. Homogenization was
carried out on arrival of the units at the Instituto de
Tecnologia de Alimentos. This institute was chosen
because of its experience and facilities in handling large
volumes of foods, equipped with different pilot plant
homogenizers suitable for different types of food matrices.
Three composite samples per product were analyzed

individually by laboratories with demonstrated laboratory capability in nutrient analysis as shown in two
interlaboratory studies using certied reference materials
from the National Institute of Standards and Technology (NIST, USA) and the Bureau of Certied
References of the European Commission (BCR). The basic
criterion for approval was a Z-score p|3|, the Z-score
being
Zscore

y  V cert
,
s

where y is the mean obtained by the laboratory, Vcert the


certied value and s the total uncertainty, the last two
values being furnished by NIST and BCR.

ARTICLE IN PRESS
R.M. Padovani et al. / Journal of Food Composition and Analysis 20 (2007) 733738

736

Table 3
Comparison of vitamin composition of common Brazilian and US foods per 100 g edible portion
NDB no.a

Description

Country

Retinol
(mcg)

Thiamin
(mg)

Riboavin
(mg)

Vit B6
(mg)

Niacin
(mg)

16014

Beans, black, mature seeds, raw

18137

Cake, white, dry mix, regular

19120

Candies, milk chocolate

08020
08506
18157

Cereals ready to eat, corn akes


Cookies, chocolate wafers

11172

Corn, sweet, yellow, canned, whole kernel, drained solids

18232

Crackers, wheat, regular

16069

Lentils, raw

20499

Macaroni, dry, unenriched

01095

Milk, canned, condensed, sweetened

01092

Milk, dry, nonfat, instant, with added vitamin A

01090

Milk, dry, whole

20038

Oats

11308

Peas, green, canned, regular pack, drained solids

20444

Rice, white, long-grain, regular, raw, unenriched

11887

Tomato products, canned, paste, with salt added

11888

Tomato products, canned, puree, with salt added

20481

Wheat our, white, all-purpose, unenriched

01116

Yogurt, plain, whole milk, 8 g protein per 8 oz

Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDAb
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA
Brazilian
USDA

0
0
0
0
o10
48
0
505
o10
3
0
0
0
0
0
0
0
0
53
73
299
709
361
253
0
0
0
0
0
0
0
0
0
0
0
0
23
27

0.117
0.900
0.177
0.235
0.050
0.112
0.727
3.3065
0.367
0.203
o0.02
0.033
0.710
0.505
0.113
0.873
0.177
0.090
0.060
0.090
0.317
0.413
0.243
0.283
0.553
0.763
0.067
0.121
0.200
0.070
o0.02
0.060
o0.02
0.025
0.313
0.120
0.037
0.029

o0.02
0.193
o0.02
0.197
0.217
0.298
1.110
2.190
0.030
0.266
0.050
0.078
0.130
0.327
o0.01
0.211
0.020
0.060
0.330
0.416
1.203
1.744
1.027
1.205
0.030
0.139
0.030
0.078
o0.02
0.049
o0.02
0.153
o0.02
0.080
o0.02
0.040
0.223
0.142

0.590
0.286
1.503
0.028
0.590
0.036
0.787
2.673
0.297
0.051
o0.02
0.047
0.170
0.136
0.417
0.540
o0.02
0.142
0.573
0.051
o0.02
0.345
o0.02
0.302
o0.02
0.119
o0.02
0.064
0.057
0.164
0.110
0.216
0.070
0.126
ndc
0.044
o0.02
0.032

4.600
1.955
1.517
1.080
0.630
0.386
10.133
22.165
1.120
2.858
3.737
1.197
7.137
4.961
5.073
2.605
3.567
1.700
o0.50
0.210
0.807
0.891
3.70
0.646
4.470
0.961
8.6
0.732
0.500
1.600
2.420
3.076
1.180
1.466
0.890
1.250
o0.30
0.075

Data taken from the Brazilian (NEPA/UNICAMP/MS, 2004) and USDA (2005) food composition tables. Values in bold represent appreciable
differences. anot determined.
a
USDA Nutrient Data Bank number.
b
The values represent the means of NDB no. 08020 and 08506.
c
ndnot determined.

The use of a Z-score p|3| as threshold for approval,


instead of the usual p|2| satisfactory score (Thompson and
Wood, 1993), is justiable because of the denominator
used. The expanded uncertainty reported in the certied
material references is the combined uncertainty times a
coverage factor k that depends on the certication
measurements distribution, whereas most laboratory or
method performance studies adopting the p|2| score use
the pooled standard deviation of the participating laboratories. The criteria would be too austere and restrictive if
the p|2| score was applied in the TACO project. The use of
expanded uncertainty is also justied by the lack of
normality for some measurements in our assays. Basic

reference for these concepts can be found at the NIST Web


site (NIST, 2003).
The USDA Food Composition DatabaseRelease 18
was chosen for the comparison (USDA, 2005). A
percentage difference between table values was calculated
for each analyte in each product, using the formula:
D% TACO value  USDA value=TACO value  100.
There are no established criteria for acceptable compositional differences. Thus, we have arbitrarily set values for
these percentage differences according to the analytes
concentrations in grams, above which the differences are

ARTICLE IN PRESS
R.M. Padovani et al. / Journal of Food Composition and Analysis 20 (2007) 733738

considered appreciable. Thus, the difference is appreciable


when it is equal to or more than (i) 10% for concentrations
X1 g, (ii) 20% for the concentration range 0.10.9 g,
(iii) 30% for the concentration range 0.010.09 g and
(iv) 40% for the concentration range 0.0010.009 g.
For proximate and mineral composition, agreement
between the Brazilian and US data for a certain product
was considered excellent when there was no appreciable
difference for all analytes. Agreement was considered very
good when appreciable difference was observed in one
analyte, good up to two analytes and fairly good for three
analytes. For vitamins, because only four or ve items were
compared, agreement was considered good when appreciable difference existed in only one analyte and fairly good
up to two analytes for those products with ve vitamins
compared.
Both the Brazilian and US data are those of pooled
samples. For corn akes, since the Brazilian values
are those of pooled, composite samples of the principal
brands and separate data are presented in the USDA
table for different brands of corn akes, the mean US
values were calculated from the data of Kellogg and
Ralston.
3. Results and discussion
Brazilian and US data on proximate, mineral and
vitamin composition of the 19 selected foods are presented
in Tables 1, 2 and 3, respectively. To facilitate the
comparison, values having appreciable differences are
represented in bold. From the tables, it can be noted that
the Brazilian data are randomly lower or greater than the

corresponding US data for any one nutrient. This is


indicative of natural variation. For analytical variability,
the Brazilian values would tend to be systematically all
lower or all higher for a certain analyte.
For the proximate composition (Table 1), agreement is
excellent for dry whole milk; good for condensed milk, dry
nonfat instant milk and canned peas; and fairly good for
black beans, milk chocolate candies, lentils, oats, white
long-grain rice and wheat our.
In terms of analyte, greater variation can be observed in
dietary ber, for which all products appeared to have
appreciable differences, and fat, for which 74% of the
products has appreciable differences (Table 4). But even for
these appreciably different foods, the results can still be
considered comparable. Considering the many factors that
can inuence the composition of foods, a 10% difference
adopted in this paper as appreciable for proximate
composition may be stringent. As reected in the means
(Table 4), the absolute differences for most of the products
considered to be appreciably different are relatively small,
even for ber and fat. Thus, for equivalent foods, data on
proximate composition may be borrowed.
Analytical uncertainty is known to increase as the
concentration of analyte goes down; thus, greater deviations due to the analysis are expected for minerals and
vitamins, adding to the overall differences in concentrations, which ideally should only be due to natural
variation. Natural variation per se would also tend to be
proportionately higher.
Very good agreement in mineral composition is noted
only in dry whole milk (Table 2). Agreement is good in
black beans and canned peas, fairly good in milk chocolate,

Table 4
Percentages and absolute differences of products with appreciable differences
Analyte

Water
Protein
Fat
Fiber
Ash
Ca
Mg
P
Fe
Na
K
Cu
Zn
Retinol
Thiamin
Riboavin
Vitamin B6
Niacin
a

% Of products with
appreciable differencesa
37
63
74
100
58
42
42
58
53
63
42
74
32
43
63
79
89
58

737

Absolute difference per 100 g edible portion of products with appreciable difference
Range

Mean

0.243.86 g
0.293.04 g
0.0810.47 g
0.413.6 g
0.100.94 g
3133 mg
858 mg
10688 mg
0.0626.67 mg
1443 mg
30334 mg
0.0090.200 mg
0.298.29 mg
48505 mg
0.0302.580 mg
0.0281.08 mg
0.0321.886 mg
0.36012.032 mg

1.86 g
1.69 g
2.02 g
2.38 g
0.40 g
33 mg
23 mg
180 mg
3.40 mg
157 mg
100 mg
0.113 mg
1.96 mg
321 mg
0.411 mg
0.213 mg
0.394 mg
3.562 mg

Number of products 15 for ber, 18 for K, 7 for retinol, 18 for vitamin B6 and 19 for the other analytes.

ARTICLE IN PRESS
738

R.M. Padovani et al. / Journal of Food Composition and Analysis 20 (2007) 733738

lentils, wheat our and natural yogurt. Results differ most


for Cu, Na and P, for which 74%, 63% and 58%,
respectively of the products evaluated appeared to have
appreciable differences (Table 4).
Even greater variation can be seen in the vitamin data
(Table 3). Agreement can be considered good only for
wheat crackers and yogurt, and fairly good for condensed
milk and dry whole milk. Variation is greater for vitamin
B6, riboavin and thiamin, with 89%, 79% and 63% of the
products, respectively, showing appreciable difference
(Table 4).
The huge differences in Fe and some vitamins in corn
akes (Tables 2 and 3) may be due to fortication of the
US products, although this was not specied in the USDA
table.
For minerals and vitamins, the absolute differences
varied widely (Table 4) and especially considering the low
levels of these analytes, the differences are in general
proportionately substantial. Thus, borrowing data for
micronutrient content should be done with caution.
Because they are micronutrients, even small differences
may be important in human health considerations.
A difculty in borrowing data is nding equivalency in
the foods of interest. For most of the foods in the Brazilian
table, equivalent foods could not be found in the USDA
tables. Meat, for example, is cut differently in Brazil; thus,
Brazilian meat cuts are not equivalent to those of the US.
For plant foods, aside from differences due to variety and
environmental conditions for those found in both countries, many fruits and vegetables widely consumed in Brazil
are not found in the US tables.

4. Concluding remarks
Borrowing data on food composition from one country to
be used in another country is limited by the lack of
equivalency in the foods consumed in these countries.
Because of genetic and environmental variation in the
composition of a given food, even in foods that could be
considered equivalent, appreciable differences in the concentrations of food components can exist, making the practice of
borrowing data questionable, especially for micronutrients.
References
Galeazzi, M.A.M., Lima, D.M., Colugnati, F.A.B., Padovani, R.,
Rodriguez-Amaya, D.B., 2002. Sampling plan for the Brazilian TACO
project. Journal of Food Composition and Analysis 15, 499505.
NEPA/UNICAMP/MS. 2004. Tabela brasileira de composic- ao de
alimentos. Versao 1. Graca e Editora Flamboyant, Campinas.
NIST Physics Laboratory. 2003. The NIST reference on constants, units
and uncertainty. Retrieved December 20, 2006 from the Physics
Laboratory Home page: /http://physics.nist.gov/cuu/UncertaintyS.
Pehrsson, P.R., Haytowitz, D.B., Holden, J.M., Perry, C.R., Beckler, D.G.,
2000. USDAs national food and nutrient analysis program: food
sampling. Journal of Food Composition and Analysis 13, 379389.
Pehrsson, P.R., Haytowitz, D.B., Holden, J.M., 2003. The USDAs
national food and nutrient analysis program: update 2002. Journal of
Food Composition and Analysis 16, 331341.
Thompson, M., Wood, R., 1993. International harmonized protocol for
prociency testing of (chemical) analytical laboratories. Journal of
AOAC International 76, 926940.
US Department of Agriculture, Agriculture Research Service, 2005.
USDA Nutrient Database for Standard Reference, Release 18.
Retrieved January 5, 2006 from the Nutrient Data Laboratory Home
page: /http://www.ars.usda.gov/ba/bhnrc/ndlS.

You might also like