You are on page 1of 7

G Model

YJFCA 2440 17
Journal of Food Composition and Analysis xxx (2014) xxxxxx

Contents lists available at ScienceDirect

Journal of Food Composition and Analysis


journal homepage: www.elsevier.com/locate/jfca
1
2
3
4
5
6

Original Research Article

Computer-generated vegan menus: The importance of food


composition database choice
Q1 Petra

Oreskovic, Jasenka Gajdos Kljusuric *, Zvonimir Satalic

Faculty of Food Technology and Biotechnology, University of Zagreb, Pierottijeva 6, 10000 Zagreb, Croatia

A R T I C L E I N F O

A B S T R A C T

Article history:
Received 5 April 2013
Received in revised form 3 March 2014
Accepted 12 July 2014
Available online xxx

Precise data on food chemical composition is crucial for any quantitative nutrition research and
indispensable for evaluation and planning of computer-based menus. Moreover, exact food chemical
composition is of the utmost importance in composing specic diets that may be low in certain essential
nutrients. Therefore, this research evaluated a weekly vegan menu using three different food
composition databases (FCDB): the Croatian, the ofcial Danish Food Composition Database and the
United States Department of Agriculture (USDA) National Nutrient Database for Standard Reference.
Signicant differences (p < 0.05) were determined in evaluated and optimised offers for the same menu
when using different FCDBs. Furthermore, in these menus quantities of some nutrients have not reached
ofcial daily recommendations needed for prevention of non-communicable diseases. In this research,
new vegan menus that can provide adequate quantity of essential nutrients, regardless of FCDB utilized,
were created with linear optimisation (LO). This resulted in high quality daily menu offers. However,
depending on the FCDB that was used, optimal menus differed in daily meal combinations, and the
effectiveness of the optimisation in the minimisation of differences caused by differences of data in
FCDBs was tested. Linear optimisation has been proved to be an effective tool in planning of specic
diets, such as vegan diets. Nonetheless, the quality of computer-planned menus depends heavily on the
quality of FCDB used.
2014 Published by Elsevier Inc.

Keywords:
Food analysis
Food composition database
Diet evaluation
Menu planning
Linear optimisation
Vegans
Vegetarians
Food composition

1. Introduction

9
10
11
12
13
14
15
16
17
18
19
20
21
22

Today we witness that due to environmental, social and health


concerns, the number of vegetarians and vegans in society is
growing. Therefore, more and more people are searching for new
diets in which they can avoid meat consumption (Kim et al., 1999).
The vegetarian initiative Meatless Monday is implemented in
Croatia as Green Monday (ZeGeVege, 2011). The initiative began
in 2003, and is promoted in Europe, Japan, the United States, Great
Britain, Canada, Israel and Australia. The basic intention was to
improve personal health and to reduce the ecological footprint of
the usual food choices (Kemp et al., 2010). Environmental issues
are one of the common reasons for choosing a vegetarian diet. The
omnivore diet requires 2.9 times more water, 2.5 times more
primary energy, 13 times more fertilizer, and 1.4 times more
pesticides than the vegetarian diet (Marlow et al., 2009).

* Corresponding author at: Faculty of Food Technology and Biotechnology,


Pierottijeva 6, HR-10000 Zagreb, Croatia. Tel.: +385 1 4605 025; fax: +385 1 4836 083.
E-mail address: jgajdos@pbf.hr (J.G. Kljusuric).

Food choices in a vegetarian diet ask for special care in dietary 23


planning because their dietary habits range from semi-vegetarians, 24
lacto-ovo vegetarians to strict vegans. Moreover, adequate 25
vegetarian diet can be appropriate for all the stages of the life 26
cycle, including pregnancy, lactation, infancy, childhood, adoles- 27
cence, as well as for athletes (ADA, 2009). Menus of these diets Q228
must be assessed and planed with additional care because the list 29
of critical nutrients in such diet is composed of proteins, n3 fatty 30
acids, vitamins: D and B12, minerals: calcium, zinc, iodine and iron. 31
It is still unclear whether choosing the strictest vegetarian diet, i.e. 32
vegan, provides extra benets in further reduction in risk of 33
chronic diseases in comparison with other vegetarians (Fraser, 34
2009; Dwyer, 1991; Craig, 2009).
35
In addition to the reasons mentioned up to now, the vegan diet 36
with its specic energy and nutrition intake is an additional 37
challenge for nutritionists and dietitians. For the evaluation of the 38
vegan diet in any quantitative study a database of food chemical 39
composition is needed. Food composition databases (FCDBs) are 40
the rst step in any quantitative study of nutrition (McCance and 41
Widdowson, 1940). When digital FCDBs are used, a large number 42
of data and information becomes accessible online, where 43
computer-generated menus can quickly and easy be further 44

http://dx.doi.org/10.1016/j.jfca.2014.07.002
0889-1575/ 2014 Published by Elsevier Inc.

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

G Model

YJFCA 2440 17
P. Oreskovic et al. / Journal of Food Composition and Analysis xxx (2014) xxxxxx

45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88

developed and evaluated (Gajdos Kljusuric et al., 2012; Korousic


Seljak, 2009). To improve the nutritional status of their populations
and to reduce the risk of chronic diseases, numerous countries
have developed their own FCDBs and nutrient-based recommendations, among them, Croatia (IOM, 2002; Chan et al., 2009; Fraser,
2009; Maillot et al., 2010; Kaic-Rak and Antonic, 1990). Ideally,
each country should have an established programme to manage its
own FCDB and to ensure that such a programme draws data from
other countries only when those values are considered applicable
to nationally consumed foods (Greeneld and Southgate, 2003).
Besides summarizing energy and nutrient data of a meal or
menu, the FCDB makes it possible to plan meals and daily meal
offers. Using optimisation in the meal-planning process will result
in balanced meal and menus that are in accordance with set
target(s). One of the most used optimisation tools is linear
optimisation (LO), which is designed to address the problem by
choosing between several possible or available variables (meals)
in order to achieve the most suitable combination (Korousic
Seljak, 2009; Maes et al., 2008). In this research, LO was used
because it allowed searching for solutions that resulted in daily
menu offers, which resolved a number of constraints, for example
recommended content of energy and nutrients in vegan diets
(Gajdos et al., 2001).
Daily nutrient requirements depend on gender, age and
physical activity level (Maillot et al., 2011). In menu planning
the aim is to minimise the percentage of persons with dietary
intakes below the Estimated Average Requirements (EAR)
(Appleby et al., 2007; Maes et al., 2008; Rumora et al., 2009).
Thus, the majority of people should fall within recommended daily
intake of nutrients or WHOs recommendations (WHO, 1991,
2003; IOM, 2002). In addition, when designing nutritionally
adequate diets for individuals, it should be possible to simultaneously take into account the special dietary needs and food
preferences of the individual (Bell et al., 2011). Applying these
premises, models were constructed in order to nd the so called
optimal solution (unpek et al., 2011; Korousic Seljak, 2009).
The objectives of this research were: (i) to specify energy and
nutrient differences in vegan diets by analysing the same diets
using different FCDBs, (ii) to use LO for daily planning of vegan
meals that are nutritionally balanced according to recommendations, combining WHO and DRI recommendations (IOM, 2001,
2002, 2004, 2011; WHO, 1991, 2003) and (iii) to identify possible
differences in the optimal results testing the effectiveness of the LO
due to the differences of data in different FCDBs.

89

2. Materials and methods

90

2.1. Dietary data

91
92
93
94
95
96
97

Vegan daily menus were downloaded in the period of 8/4/2011


till 2/9/2011 from the website of the Croatian vegetarian society
Animal friends that are presented as a part of international
initiative Meatless Monday which is also implemented in Croatia
(ZeGeVege, 2011). The daily menu consisted of three meals
(breakfast, lunch and dinner), and one snack. Seven menus were
randomly chosen to create one weekly menu (Table 1).

98

2.2. Daily energy and nutrient needs

99
100
101
102
103
104

Anthropometry (body weight and body high) and age, along


with gender, should be taken into account when evaluating daily
energy and nutrient needs. According to gender, a vegetarian is
more likely to be female, and according to age, a young adult
(Phillips, 2005). Therefore, as a reference subject a female vegan,
22 years old with body weight (59 kg) and body height (169 cm)

Table 1
Weekly vegan menu.
Meal

Abbreviation

Meal
components

Breakfasts

B1
B2
B3
B4

Oatmeal
Pasta with olive paste, fresh orange juice, apples
Oatmeal with bananas and grapefruit
Milkshake with strawberries, bananas and walnuts,
pasta
Multi-grain bread with margarine and marmalade,
grapefruit juice
Polenta with sweet syrup and almond decoration
Muesli with nuts, raisins and soymilk

B5
B6
B7
Lunches

L1

Morocco soup with chickpeas, millet with tofu and


asparagus, Orange and banana juice
Minestrone soup, spring spinach dumplings, potato
salad
Peas soup, rice crockets, artichokes with potatoes
and spring onions, poppy seeds cake
Chickpeas with vegetables, beets salad, bread,
American apple pie
Autumn cream pumpkin soup, peppers lled with
potatoes, carrot salad, pears covered with chocolate
Stew with vegetables and legumes, broccoli
crockets, lettuce, banana beads
Minestrone soup, baked peppers lled with
tomatoes, onions and parsley, potato salad with
chicory and tofu

L2
L3
L4
L5
L6
L7

Suppers

S1
S2
S3
S4

Pickled eggplants and squashes


Mushroom mix from wok
Chicory salad with oranges
Polenta with vegetables and mushrooms, broccoli
salad, bread
Lettuce with corn, buckwheat cake
Squash lled with mushrooms
Rice with lentils and lemon

S5
S6
S7
Snacks

SN
SN
SN
SN
SN
SN
SN

Fruits

F 1F 5
F 6F 11

1
2
3
4
5
6
7

Sunower seeds, blueberries, plums


Raisins, almonds, pears
Strawberries, pears, walnuts, apricots, blueberries
Blackberries, apricots, dried gs, almonds, grapes
Grapes, apples, almonds, peanuts, blackberries
Plums, almonds, walnuts, bananas
Peaches, walnuts, almonds, blueberries, raspberries
Bananas, apples, orange, pears, tangerines
Almonds, raisins, walnuts, apricots, gs, peanuts

was chosen, dened according to a previous Croatian study among


young adults (Colic Baric et al., 2003).
The estimated energy requirement (EER) was calculated using the
recommendations (IOM, 2002).Vegetarians take more exercise than
comparable non-vegetarians (Key et al., 1999), so as the physical
activity coefcient was used the one that is proposed for active
persons (1.27), implying that the referent subject exercises regularly,
so the total daily energy expenditure is 9980 kJ (2384 kcal).

105
106
107
108
109
110
111
112

2.3. Energy and nutrient analysis of vegan menus

113

The menus were analysed using three different FCDBs:

114

 Croatian FCDB (Kaic-Rak and Antonic, 1990)


 Danish FCDB, FDB7.01 (DTU, 2008)
 USDA FCDB, rel.23 (USDA, 2011)

117
116
118
119
120
121

The USDA National Nutrient Database for Standard Reference


and Danish Food Composition Databank are examples of FCDBs
that are freely available in the public domain with their own
websites.
Evaluation entailed summarizing energy and nutrient content,
multiplied by the portion size of raw ingredients that are
components of an offer (meal or menu). Analysis of an offer

122
123
124
125
126
127
128

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

G Model

YJFCA 2440 17
P. Oreskovic et al. / Journal of Food Composition and Analysis xxx (2014) xxxxxx

129
130
131
132
133
134
135
136
137
138
139
140

compliant with the recommendations of IOM (2002) and WHO


(2003) was conducted on a daily basis for each FCDB and presented
as average deviation. For vegetarians, estimated iron requirements
and recommended intakes are increased by a factor of 1.8 to
account for the reduced absorption, and the requirements for
dietary zinc are increased by a factor of 1.5, particularly for vegans
who, due to high consumption of grains and legumes, ingest
sample amounts of phytate (IOM, 2001). For the recommended
intake of calcium recently revised DRI were used (IOM, 2011). The
main features of the FCDBs that were used are presented in Table 2.
The standard recipe calculation procedure has been followed as
suggested by Reinivuo et al. (2009).

141

2.4. Linear optimisation

142
143
144

Basic structure of the linear model is subjecting the goal


function with constrains that will restrict the energy and nutrient
content as follows:

145
Min F w j  B j w j  L j w j  S j w j  SN j wk  F k
148
147
146

Subject to

149
aij  B j aij  L j aij  S j aij  SN j w j  F j < or or > bi
152
151
150
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172

Where: B Breakfast; L Lunch; S Supper; SN Snack; F


Fruits; j, k number of the meals (j), j = 1, . . ., 7; k = 1, . . ., 11; aij
Content of energy, water or nutrients, i, i = 1, 2,. . ., 20, for observed
meals, j; bi daily recommended intakes of energy, water or
nutrients; wj the weight factor that includes the case based and
rule based reasoning for offered food choices.
The optimisation was conducted using recommendations given
by the Food and Nutrition Board of the Institute of Medicine as the
DRIs, which were combined with recommendations given by
WHO; this is presented in Table 2. Since vegans have no dietary
source of vitamin B12, we have not included this nutrient in the
optimisation process. As vitamin D in vegetarian diets is not
decisive, and there are only few dietary sources and these are of
animal origin (Chan et al., 2009), we have also omitted vitamin D
from the optimisation results.
To dene the daily recommended energy and nutrient intakes,
the DRIs and WHO recommendations were used (bi the right side
of constrains that are subjected to the goal function). Studies of
energy and nutrient daily offers (Rumora et al., 2009; unpek et al.,
2011) showed the necessity of partial deviation (10%) for RDA
limitations given by Food and Nutrition Board, Institute of Medicine
Table 2

Q7 Daily dietary recommendations from the IOM (2002) and WHO (2003) used as
limitations in the linear optimisation.
Observed
limitation

Energy (Ed)
Protein
Dietary bre
Carbohydrates
Total fat
Saturated fat
Sodium
Calcium
Iron
Zinc

DRI

WHO

Limitations
according to used
food databases

Lower
limitation

Table 3
Basic features of used food composition databases (FCDBs).
Feature of the FCDB

Croatian

Danish, FCDB7.01

USDA, rel. 23

Year of publication
Number of items
Number of components
Include processed
food/prepared meals

1991
582
34
Yes/No

2008
1049
113
Yes/Yes

2010
7636
34
Yes/Yes

(IOM, 2003). When the WHO recommendations were used, lower and
upper limitation is already dened, and they were used as bi.

173
174

2.5. Statistical analysis

175

In order to deduct the critical points regarding observed daily


offers calculated using different FCDBs and selected constrains, the
sensitivity test in the LINDO program was used. To detect
signicantly differences in daily offers, one-way ANOVA was used.
To identify the cases that differ, the principal component analysis
was applied in the program Statistica v.10. (StatSoft, 2011, Tulsa,
OK, USA).

176
177
178
179
180
181
182

3. Results and discussion

183

Differences of the FCDBs (Table 3) are large concerning the


number of items, components to inclusion (or exclusion) of
processed food. In the Croatian FCDB spices that are commonly
used in vegetarian diets (like cumin and basil) are not listed, nor
are different fat sources like palm oil, or whole grain bread or dried
tomatoes. The gaps and obsolescence of the Croatian FCDB conrm
the necessity of updating and extending it. Besides the number of
foods, the FCDBs show differences regarding the method of
preparation of foods. The USDA FCDB provides information for the
raw, but also for a processed food (boiled, baked, dried, drained,
etc.), but in the Danish FCDB only a small part of listed food is
thermally processed, and the Croatian FCDB contains data just for
raw foods. Therefore to be able to compare the results, calculation
was conducted using the same ingredients in their raw form in all
three FCDBs.
Weekly menus (Table 1) were evaluated using the same food
ingredients in different FCDBs, and the average energy and nutrient
contents are presented in Table 4.
The analysed offers differ in the content of energy, which
ranged from 8960 to 10250 kJ. The signicant difference in the
average energy content (p = 0.03) is a result of signicant
differences in the content of carbohydrates (p = 0.023), which
ranged from 295 g/day (Croatian FCDB) to almost 350 g/day (USDA
FCDB).

184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207

Table 4
Comparison of average energy and nutrient content for the 7-day vegan menu
calculated using data from three different FCDBs (mean and SD) for the same food
ingredients.
Observed content

8400 kJ < Ed < 10,255 kJ


>41 g
>25 g
>250 g
<90 g
<25 g
1500 mg < Na < 2300 mg
>720 mg
>22.5 mg
>9 mg

10% Ed
27 g
55% Ed
15% Ed
0% Ed
/

Upper
limitation

15% Ed
40 g
75% Ed
30% Ed1
10% Ed
6g

FCDBs
Danish

Croatian

Energy (kJ)
Protein (g)
Dietary bre (g)
Carbohydrates (g)
Total fat (g)
Saturated fats (g)
Sodium (mg)
Calcium (mg)
Iron (mg)
Zinc (mg)

USDA

SD

SD

SD

8959.4
57.5
64.0
295.3
88.6
12.1
2252.7
685.3
22.3
8.6

749.9
6.1
5.9
26.8
11.9
1.6
262.5
69.6
2.1
1.1

9486.1
62.0
55.2
341.0
88.5
16.3
1771.5
1079.1
24.1
8.7

935.7
6.9
4.7
31.9
13.7
3.2
241.7
177.1
4.1
1.1

10252.3
62.6
64.4
347.3
90.6
11.9
2057.3
1092.3
28.5
12.7

843.8
6.1
5.4
30.7
12.9
1.8
188.5
163.6
4.4
1.7

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

G Model

YJFCA 2440 17
4

P. Oreskovic et al. / Journal of Food Composition and Analysis xxx (2014) xxxxxx

208
The Reinivuo method (Reinivuo et al., 2009) includes the
209
calculation of energy and nutritional composition by the multipli210
cation of the content with appropriate yield and retention factors,
211
but this method was not applied in this study because the factors
212
were not known for all FCDBs. If the same yield and retention
213
factors had been used for different FCDBs, the effect would have
214
been the same as the effect that is achieved when the energy and
215
nutrient content for all meal ingredients were just multiplied with
216
their content and summarized. A study conducted in 1999 by
217
McCollough and co-workers compared four nutritional databases
218
in the DASH trial where the results showed deviations that varied
219
for most nutrients due to the chemical analysis used in detection of
220
the food components.
221
Critical nutrients for a vegetarian diet according to the
222
American Dietetic Association (2009) include proteins, n3 fatty
223
acids, iron, zinc, iodine, calcium, and vitamins D and B12. Vitamin
224
B12 intake among Croatian vegetarians and vegans, based on
225
weighed 7-days food records, is estimated at 1.6 and 0.2 mg,
226
respectively (Satalic, 2009). Due to limited food choices, additional
227
effort is needed to plan an adequate vegan diet. Beside the daily
228
energy requirements for vegans, requirements for nutrients that
229
were available in the FCDBs and were considered as important for
230
the target group, as proteins, dietary bres, total carbohydrates,
231
total fat, saturated fats, calcium, sodium, zinc, iodine and iron were
232
observed. Being a vegan implies a lower quality regarding protein
233
intake and especially the amino acid lysine (Kniskern and Johnston,
234
2011), and the recommended daily intake rises for vegans from
235
recommended 0.8 for general population to 1 g/kg body weight.
236
Using one-way ANOVA, insignicant deviations (p > 0.05) were
237
established for the content of proteins, fats, poly-unsaturated fats
238
and iron. Signicant differences (p < 0.05) were observed for the
239
content of sodium (where the lowest content was achieved by use
240
of Danish FCDB), content of calcium and zinc (where the lowest
241
content was achieved using the Croatian FCDB). Mentioned
242
differences can be explained by use of different denition of the
243
same food because each country is using their country-specic
244 Q3 names (Deherveng et al., 1999), using different (allowed) amounts
245
of fertilizer, in different countries (Hakala et al., 2003) and the
246
geographic locations of the food sampling can also cause the
247
differences (Schulze, 1999; Uusitalo et al., 2011) as well as use of
248
different methods in food component detection which is in our
249
opinion the most likely cause (Schulze, 1999; Uusitalo et al., 2011).
250
Analysing the absolute deviations from the recommendations,
251
deviations for some nutrients were determined (p < 0.05), such as
252
the deviations for saturated fats that are in accordance with the
253
share of 10% of the daily energy intake, but are not in the
254
recommended range regarding the DRI limitation (Table 2).
255
Also, dietary bre does not meet the recommended WHO range

of 2740 g but is in accordance with the DRIs (>27 g) because it


contains only the recommendation for the minimal daily intake.
High content of dietary bre in the meals is a result of notable
intakes of vegetables, fruits and nuts, which is also reected on
the content of proteins, which is the secondary outcome of legume
intake.
Danish and USDA FCDBs are mostly in the range of identied
1015% differences (McCullough et al., 1999) with the exception of
iron, but the nutrient content in foods in the Croatian FCDB differs
notably for carbohydrates, fats, and observed minerals, which was
also proven with ANOVA (p < 0.05). In the study of Uusitalo et al.
(2011), the between-country comparisons of nutrient data were
presented and the content of carbohydrates differed as an outcome
of measurements and analysis methods.
In order to determine which FCDB most resembles another, Q4
principle component analyses were implemented; results are
presented in Fig. 1.
Analysing results of observed content of energy and nutrients
(Table 4), it can be seen that for some nutrients the same vegan
menu will vary extremely, such as for calcium (min. value Croatian
FCDB, max. value USDA). This is a problem for multivariate analysis
because we have a matrix consisting of three different databases
with associated cases (energy and nutrient content). So, the
principle component analysis (Fig. 1a) and corresponding loadings Q5
(Fig. 1b) show that the Croatian FCDB is more like the USDA food
database because the factor l is explaining the majority of the
variation between the observed variables (63.27%). On the other
hand, factor 2 includes the variation of 36.73% in the data matrix, so
this is an argument why in a data analysis in Croatia the USDA FCDB
could be used, beside the known fact that frequent updates of the
FCDBs are important (Uusitalo et al., 2011). In presented loadings of
the principal component analysis (Fig. 1b), the second quadrant
of the correlation circle shows those cases that differ extremely
using the Croatian FCDB (positioned in the fourth quadrant, Fig. 1a)
content of proteins, carbohydrates and iron. Comparing this nding
with the average values of mentioned nutrients in Table 4, it is
obvious that when the Croatian FCDB is used this resulted with the
minimal content of mentioned nutrients.
The contents of the observed minerals (Na, Fe, Ca and Zn) are
notably different, especially comparing the FCDB of Croatia and the
data given in FCDBs of Denmark and USDA (Fig. 1b). According to
Uusitalo et al. (2011), as a measurement and analytical method for
iron, zinc and calcium, atomic absorption spectrometry is used
nowadays and it is also expected to be a measurement tool for
sodium, but this method was not available in the 1990s when the
Croatian FCDB was published, another argument that supports the
need for renewal of this FCDB or to use some FCDB from the region,
if available (e.g. from the Czech Republic).

Fig. 1. Multivariate analysis of the average daily energy and nutrient content of vegan menus using different FCDBs presented with (a) rst two factors of the principle
component analysis and (b) associated loadings in the correlation circle.

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303

G Model

YJFCA 2440 17
P. Oreskovic et al. / Journal of Food Composition and Analysis xxx (2014) xxxxxx

304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347

348
349
350
351
352
353
354
355
356
357
358

Analysing and comparing data values for specic foods resulted


in differences for some observed parameters that are far from an
acceptable coefcient of variation (CV = 10%). One such example
was the energy content for crude wheat bran where 100 g in the
USDA FCDBs (wheat bran, crude) has 1500 kJ, in the Danish FCDB
(wheat bran) has 1216 kJ and in the Croatian FCDB the same food
contains 40% less energy (862 kJ). Large differences in the energy
content were also detected for raw tomatoes. The Danish FCDB
contains information that 100 g of fresh tomatoes (tomato, ripe,
raw, unknown origin) contains 107 kJ of energy, and in the
Croatian FCDB (tomato, red), the content of energy is just one half
of it, 60 kJ.
After the analysis of the vegan meal offers and determination of
average insufcient or excessive daily intake, the next step is an
intervention in the meal offers using computer generated menu
planning (Sterling et al., 1996; Maillot et al., 2011). Optimisation in
meal planning is the process of searching the best, i.e. the optimal
daily meal offer(s) (Gajdos Kljusuric et al., 2012) that include(s) a
variety of choices and recommendations that could be added in the
menu planning with the main goal, namely meeting the individual
needs (Sterling et al., 1996; Gaal et al., 2007), and it can be used in
health education or in individual adaptation to a new diet, known
as computer-tailoring (Maes et al., 2008). In LO, the goal function is
mostly the price of meals subjected to constrains that dene the
required maximum or minimum intakes of observed nutrients
(unpek et al., 2011), but in this study the goal function was the
minimisation of meal combinations that include the case-based
and rule-based reasoning (CBR & RBR). Minimisation that includes
CBR & RBR should result in different meal offers, where the meals
differ in texture, colour, method of preparation, and allowing
trade-offs (Marling et al., 1999). The used tool in the intervention is
linear programming because it is the most used tool in computer
menu planning (Maillot and Drewnowski, 2011; unpek et al.,
2011; Gaal et al., 2007; Maillot et al., 2011, 2010; Maes et al., 2008).
One of the aims was to use the meals given in Table 1 and to
improve them by adding some new food. In the menu planning a
fruit group was added, listed in Table 1 as F6F11. The purpose of
adding fruits in the daily offer was to try to meet daily nutrient
recommendations, and LO was used as a tool that can nd optimal
offers regardless of the number of added variables or constrains
(Maillot et al., 2011). For the weekly menu (Table 1), where each
daily offer consists of breakfast, lunch, dinner, snack and fruit,
72,030 combinations can be obtained (7B  7L  7S  7SN  5F1
5  6F611). But concerning the energy and nutritional quality of
each daily offer, the number of possible combinations will be
signicantly reduced because some of the daily menus will not
meet the daily energy and nutrient requirements (unpek et al.,
2011). However, the analysis already shows that recommended
menu by the Meatless Monday initiative could mislead
consumers making them believe that they know what they offer.
Computer-generated menu offers are an objective approach.
Optimisation was performed separately for each data set obtained
from the Croatian, Danish and the USDA FCDB. Each new daily
vegan menu consisted of one breakfast, lunch, supper, snack and
additionally a portion of fruits.

Table 5
Optimal daily offer using different FCDBs.

Croatian FCDB
Danish FCDB
USDA FCDB

Breakfast
(B)

Lunch
(L)

Supper
(S)

Snack
(SN)

Fresh
fruits
(F15)

Seeds and
dried fruits
(F611)

6
5
7

1
4
4

1
5
1

6
3
2

1
4
4

7
11
8

In accordance with the ndings of McCullough et al. (1999) is


the outcome that the use of different FCDBs resulted in different
optimal solutions (Table 5) where the nutrient intakes were in
those ranges given in Table 2. In WHO recommendations (WHO,
1991), the lower limit for a nutrient is the minimal amount of it
that should prevent disease caused by lack of the nutrient, and the
upper limitation is the ceiling of the maximal intake, also set in
order to prevent chronic disease (IOM, 2004).
In order to test the optimisation efcacy in the minimisation of
differences for the energy and nutrient contents, the same optimal
offer was evaluated applying different FCDBs. In some studies LO is
applied to harmonize an excessive intake of nutrients with the
recommendations (Maillot and Drewnowski, 2011; unpek et al.,
2011). When it is applied for vegan menu plans, it is necessary to
use LO in order to raise the intake of certain foods that are rich on
those nutrients considered to be insufcient in their nourishment
(Kniskern and Johnston, 2011). The idea was to embed it in the
optimisation menu planning in order to meet the daily nutrient
recommendations (Gaal et al., 2007; Maillot et al., 2011; Maillot
and Drewnowski, 2011; Rumora et al., 2009). And this was
successful achieved for the vegan menus. In Table 6 are
summarised means of nutrient contents for the same daily menu
offer resulting from the optimisation conducted using three
different FCDBs, when the daily offer consisted of following meals
(Table 5): breakfast, B6; lunch, L1; supper, S1; snack, SN3; and
fruits: F4 and F8. The results again showed a variation in the
nutrient contents, but these variations were reduced regarding
minimal and maximal limitations of the LO (from Table 2), except
saturated fats that are the highest when the Croatian FCDB was
used (24.9 g).
The sensitivity analysis, in this case, is used to detect critical
nutrients that are mostly responsible for why a daily offer will fail
from a large set of possible daily offers (72,030), and it is so because
it cannot reach the minimal restriction (Table 2). Sensitivity
analysis is conducted after the optimisation, and results in
different critical nutrients that depend on which FCDB was used.
For the Croatian FCDB, the set of critical nutrients is the largest one
and includes content of dietary bres, carbohydrates, total fats,
saturated fats and Calcium. The smallest set of critical nutrients
was observed when the USDA FCDB was used. This set was limited
to the content of calcium and total fats. Detection of critical points
is important to dene the cause(s) if a combination of meals will
not build an optimal daily offer.
The optimisation results also pointed out that the computergenerated optimal daily menu is not a daily meal combination
(according to Table 1) recommended by the Meatless Monday
initiative, which is an indication that those consumers who trust
some initiatives (as mentioned) can be misled.
Each evaluation is inuenced by the quality of initial data, but
the optimisation used reduced the differences of average daily
Table 6
Average content of energy and nutrients of an optimised menu for vegans.
Observed content

FCDBs
Danish

Croatian

Energy (kJ)
Protein (g)
Dietary bre (g)
Carbohydrates (g)
Total fat (g)
Saturated fats (g)
Sodium (mg)
Calcium (mg)
Iron (mg)
Zinc (mg)

USDA

SD

SD

SD

9158.2
56.6
62.6
279.5
73.2
24.9
2215.6
797.9
22.4
8.7

904.1
13.7
12.8
40.5
24.9
4.8
430.8
130.7
3.6
0.8

9340.5
58.5
58.7
334.5
86.1
10.2
1909.2
917.2
18.9
8.8

758.4
10.6
5.2
53.0
16.5
3.0
621.0
183.8
3.4
0.9

9597.1
59.6
64.2
331.2
81.2
10.9
2106.3
989.8
27.2
10.9

799.4
9.5
4.0
48.3
17.5
1.7
498.4
150.9
6.1
1.8

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408

G Model

YJFCA 2440 17
P. Oreskovic et al. / Journal of Food Composition and Analysis xxx (2014) xxxxxx

409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426

offer from the recommendation (10.3%) for all nutrients except


saturated fats and iron, which showed variations, and the
established signicant differences (p < 0.05) in the nutritional
composition of optimal food choices, which is caused by signicantly different values for the same foods in the analysed FCDBs.
Computer-generated menus for vegans using different FCDBs
have shown that computer-tailoring for a specic population or an
individual as a vegetarian or vegan is possible. Application of the
computer in menu planning is an objective tool, but the interpretation of achieved results and the goals that were set have shown that
the nal results, i.e. the optimal menu offers, are inuenced by the
quality of the input data, namely the FCDB that was used.
The computer menu analysis, planning and optimisation is
more effective if the FCDB contains a variety of foods with updated
content of nutrients that are a result of application of new
measurements and analytical methods as well as offering new
foods, following the supply on the market, like the FCDBs of
Denmark and USDA.

427

4. Conclusions

428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465

Croatian, Danish, and USDA FCDBs differed with respect to the


number of food items that are included, especially those that are
usually consumed by vegans, as well as to the extent of
information given for each food. The advantage of the Danish
and USDA FCDBs was the inclusion of information about processed
food, which facilitated the menu evaluation and planning. This
implied that the evaluation was inuenced by the quality of initial
data. The chosen FCDB should include foods consumed by the
target group (e.g. vegans) as well as the nutrients that should be
observed in the evaluation or/and optimisation process.
Analysis of the weekly menu proposed by a vegetarian society
resulted in different average energy and nutritional content of the
observed menus, which have not met the daily dietary recommendations demonstrating the necessity of intervention, if such a
diet is planned for a long time. As a useful intervention tool, LO was
used, generating menus using the same 7-days vegan menu plans
but enriched with fruits. Computer-generated menu plans have
resulted in different daily menu combinations i.e. optimal menus,
as a consequence of different input data in this case, different
FCDBs. Differences were signicant (p < 0.05) for carbohydrates,
saturated fats, calcium, sodium, zinc and iron. Other nutrients were
in the range of 15% deviation, in comparison to what other studies
conrmed as acceptable, which indicates the successful applicability of
LO in minimisation of the differences of FCDBs. Such optimised menu
offers could also be used for education of new vegans in order to
indicate the importance of dried fruits and nuts that would increase the
intake of critical nutrients. And nally, LO, as a meal planning tool, has
shown its successful applicability in the computer generation of vegan
menus (planning and/or optimisation). The question still remains,
however, of the quality of the FCDB, and this cannot be avoided with an
objective tool in meal planning, as is the case with optimisation.
Frequent updates of the FCDBs are necessary and should reect
the food offered on the market. In cases where the updates are
uncommon, as in Croatia, it would be advisable to use the FCDB
from a similar region. So, comparing three different FCDBs, the one
that should be used for vegan menus evaluation and planning in
Croatia, is the USDA FCDB, which corroborates with the results of
the principle component analysis.

466 Q6 Uncited Reference


467
468
469

Ahuja and Perloff (2008), Cunningham and Sobolewski (2011),


Glibetic et al. (2011), Harrison (2004), Simila et al. (2006) and
Yamini et al. (2006).

References

470

Ahuja, J.K.C., Perloff, B.P., 2008. Quality control procedures for the USDA Food and
Nutrient Database for Dietary Studies nutrient values. Journal of Food Composition and Analysis 21, S119S124.
Appleby, P., Roddam, A., Allen, N., Key, T., 2007. Comparative fracture risk in
vegetarians and nonvegetarians in EPIC-Oxford. European Journal of Clinical
Nutrition 61, 14001406.
Bell, S., Colombani, P.C., Pakkala, H., Christensen, T., Mller, A., Finglas, P.M., 2011.
Food composition data: identifying new uses, approaching new users. Journal of
Food Composition and Analysis 24, 727731.
Colic Baric, I., Satalic, Z., Lukesic, Z., 2003. Nutritive value of meals, dietary habits
and nutritive status in Croatian university students according to gender.
International Journal of Food Science and Nutrition 54, 473484.
Chan, J., Jaceldo-Siegl, K., Fraser, G.E., 2009. Serum 25-hydroxyvitamin D status of
vegetarians, partial vegetarians, and nonvegetarians: the Adventist Health
Study. American Journal of Clinical Nutrition 89, 1686S1692S.
Craig, W.J., 2009. Health effects of vegan diets. American Journal of Clinical Nutrition 89, 627S1633S.
Cunningham, J., Sobolewski, R., 2011. for nutrition labelling: experience from
Australia. Journal of Food Composition and Analysis 24, 682685.
DTU, 2008. Fodevaredatabanken, Version 7.01. DTU-Danmarks Tekniske Universitet
Retrived April 15, 2011 from:http://www.foodcomp.dk.
acic, J., Kurtanjek, Z., 2011.
unpek, S., Gajdos Kljusuric, J., Magdic, D., Lukinac C
Optimisation of the daily nutrient composition of daily intakes during gestation.
Croatian Journal of Food Technology, Biotechnology and Nutrition 6, 4551.
Dwyer, J.T., 1991. Nutritional consequences of vegetarianism. Annual Review of
Nutrition 11, 6191.
Fraser, G.E., 2009. Vegetarian diets: what do we know of their effects on common
chronic diseases? American Journal of Clinical Nutrition 89, 1607S1612S.
Gaal, B., Vassanyi, I., Kozmann, G., 2007. Application of articial intelligence for
weekly dietary menu planning. Studies in Computational Intelligence 65,
2748.
Gajdos Kljusuric, J., Rumora, I., Kurtanjek, Z., 2012. Application of Fuzzy Logic in diet
therapy advantages of application. In: Dadios, E.P. (Ed.), Fuzzy Logic
Emerging Technologies and Applications. INTECH, Rijeka, Croatia.
Gajdos, J., Vidacek, S., Kurtanjek, Z., 2001. Meal planning in boarding schools in
Croatia using optimisation of food components. Current Studies of Biotechnology 2, 217222.
Glibetic, M., Kadvan, A., Tepsic, J., Debeljak Martacic, J., Djekic-Ivankovic, M.,
Gurinovic, M., 2011. Management of food composition database harmonized
with EuroFIR criteria using a web application. Journal of Food Composition and
Analysis 24, 741743.
Greeneld, H., Southgate, D.A.T., 2003. Food Composition Data: Production, Management and Use, 2nd ed. FAO, Rome.
Hakala, P., Knuts, L.R., Vuorinen, A., Hammar, N., Becker, W., 2003. Comparison of
nutrient intake data calculated on the basis of two different databases: results
and experiences from a Swedish-Finnish study. European Journal of Clinical
Nutrition 57, 10351044.
Harrison, G.G., 2004. Fostering data quality in food composition databases: applications and implications for public health. Journal of Food Composition and
Analysis 17, 259265.
IOM, Institute of Medicine, Food and Nutrition Board, 2001. Dietary Reference
Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine,
Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc. National
Academy Press, Washington, DC.
IOM, Institute of Medicine, Food and Nutrition Board, 2002. Dietary Reference
Intakes for Energy, Carbohydrates, Fiber, Fat, Protein and Amino Acids (Macronutrients). National Academy Press, Washington, DC.
IOM, Institute of Medicine, Food and Nutrition Board, 2003. Institute of Medicine,
Food and Nutrition Board, 2004. Dietary Reference Intakes Application in
Dietary Planning. National Academy Press, Washington, DC.
IOM, Institute of Medicine, Food and Nutrition Board, 2004. Institute of Medicine,
Food and Nutrition Board, 2004. Dietary Reference Intakes for Water, Potassium, Sodium, Chloride, and Sulfate. National Academy Press, Washington, DC.
IOM, Institute of Medicine, Food and Nutrition Board, 2011. Dietary Reference
Intakes for Calcium and Vitamin D. National Academy of Sciences Press,
Washington, DC.
Kaic-Rak, A., Antonic, K., 1990. Tablice o sastavu namirnica i pica. Zavod za zastitu
zdravlja SR Hrvatske, Zagreb.
Kemp, K., Insch, A., Holdsworth, D.K., Knight, J.G., 2010. Food miles: do UK consumers actually care? Food Policy 35, 504513.
Key, T.J., Fraser, G.E., Thorogood, M., Appleby, P.N., Beral, V., Reeves, G., Burr, M.L.,
Chang-Claude, J., Frentzel-Beyme, R., Kuzma, J.W., McPherson, K., 1999. Mortality in vegetarians and non-vegetarians: detailed ndings from a collaborative
analysis of ve prospective studies. American Journal of Clinical Nutrition 70,
516S524S.
Kim, E.H.J., Schroeder, K.M., Houser, R.F., Dwyer, J.T., 1999. Two small surveys,
25 years apart, investigating motivation of dietary choice in 2 groups of
vegetarians in the Boston area. Journal of the American Dietetic Association
99, 598601.
Kniskern, M.A., Johnston, C.S., 2011. Protein dietary reference intakes may be
inadequate for vegetarians if low amounts of animal protein are consumed.
Nutrition 27, 727730.
Korousic Seljak, B., 2009. Computer-based dietary menu planning. Journal of Food
Composition and Analysis 22, 414420.

471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

G Model

YJFCA 2440 17
P. Oreskovic et al. / Journal of Food Composition and Analysis xxx (2014) xxxxxx

555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588

Maes, L., Vereecken, C.A., Gedrich, K., Rieken, K., Sichert-Hellert, W., De Bourdeaudhuij, I., Kersting, M., Manios, Y., Plada, M., Hagstromer, M., Dietrich, S., Matthys,
C., on behalf of the Study Group HELENA, 2008. A feasibility study of using a diet
optimization approach in a web-based computer-tailoring intervention for
adolescents. International Journal of Obesity 32, S76S81.
Maillot, M., Drewnowski, A., 2011. Energy allowances for solid fats and added sugars
in nutritionally adequate U.S. diets estimated at 1733% by Linear Programming
Model. Journal of Nutrition 14, 333340.
Maillot, M., Issa, C., Vieux, F., Lairon, D., Darmon, N., 2011. The shortest way to reach
nutritional goals is to adopt Mediterranean food choices: evidence from computergenerated personalized diets. American Journal of Clinical Nutrition 94, 11271137.
Maillot, M., Vieux, F., Amiot, M.J., Darmon, N., 2010. Individual diet modeling
translates nutrient recommendations into realistic and individual-specic food
choices. American Journal of Clinical Nutrition 91, 421430.
Marling, C.R., Petot, G.J., Sterling, L.S., 1999. Integrating case-based and rule-based
reasoning to meet multiple design constraints. Computational Intelligence 15,
308332.
Marlow, H.J., Hayes, W.K., Soret, S., Carter, R.L., Schwab, E.R., Sabate, J., 2009. Diet
and the environment: does what you eat matter? American Journal of Clinical
Nutrition 89, 1699S1703S.
McCance, R.A., Widdowson, E.M., 1940. The chemical composition of foods. Medical
Research Council Special Report Series No. 235, His Majestys Stationery
Ofce, London.
McCullough, M.L., Karanja, N.J.M., Lin, P.-H., Obarzanek, E., Phillips, K.M., Laws, R.L.,
Vollmer, W.M., OConnor, E.A., Champagne, C.M., Windhauser, M.M., 1999.
Comparison of 4 nutrient databases with chemical composition data from
the Dietary Approaches to Stop Hypertension trial. Journal of the American
Dietetic Association 99, S45S53.
Phillips, F., 2005. Vegetarian nutrition. Nutrition Bulletin 30, 132167.
American Dietetic Association, 2009. Position of the American Dietetic Association:
vegetarian diets. Journal of the American Dietetic Association 109, 12661282.
Reinivuo, H., Bell, S., Ovaskainen, M.-L., 2009. Harmonisation of recipe calculation
procedures in European food composition databases. Journal of Food Composition and Analysis 22, 410413.

Rumora, I., Gajdos Kljusuric, J., Bosanac, V., 2009. Analysis and optimisation of
calcium content in menus and dairy offer in Croatian kindergartens. Mljekarstvo 53, 201208.
Schulze, A., 1999. Verglaich der Ernahrungssituation in Deutschland und Grobrtanien. Peter Lang GmbH, Frankfurt a/M, Germany.
Simila, M., Ovaskainen, M-L., Virtanen, M.J., Valsta, L.M., 2006. Nutrient content
patterns of Finnish foods in a food composition database. Journal of Food
Composition and Analysis 19, 217224.
Sterling, L., Petot, G., Marling, C., Kovacic, K., Ernst, G., 1996. The role of common
sense knowledge in menu planning. Expert Systems with Applications 11,
301308.
Satalic, Z., 2009. Vegetarian Diet, B Vitamins Nutritional Status and Homocysteine
as Risk Factors for Osteoporosis. Faculty of Food Technology and Biotechnology,
University of Zagreb, Zagreb, Croatia PhD Thesis.
USDA, U.S. Department of Agriculture, Agricultural Research Service, 2011. USDA
National Nutrient Database for Standard Reference Release 24. Online: Retrieved May 11, 2011 from:http://www.ars.usda.gov/ba/bhnrc/ndl
Uusitalo, U., Kronberg-Kippila, C., Aronsson, C.A., Schakel, S., Schoen, S., Mattisson, I.,
Reinivuo, H., Silvis, K., Sichert-Hellert, W., Stevens, M., Norris, J.M., Virtanen,
S.M., The TEDDY Study Group, 2011. Food composition database harmonization
for between-country comparisons of nutrient data in the TEDDY Study. Journal
of Food Composition and Analysis 24, 494505.
WHO, World Health Organisation, 1991. Diet, nutrition and the prevention of
chronic diseases a report of the WHO Study Group on Diet, Nutrition and
Prevention of Noncommunicable Diseases. Nutrition Reviews 49, 291301.
WHO, World Health Organisation, 2003. Diet, nutrition and the prevention of
chronic diseases Report of the Joint WHO/FAO Expert Consultation. WHO
Technical Report Series, No. 916. (TRS 916), .
Yamini, S., Juan, W.Y., Marcoe, K., Britten, P., 2006. Impact of using updated food
consumption and composition data on selected MyPiramid Food Group
nutrient proles. Journal of Nutrition Education and Behaviour 38,
S142S316.
ZeGeVege, 2011. Green Monday Vegetarian Menus. Retrieved between 8/4/20112/9/2011, from: http://www.zeleni-ponedjeljak.com.

Please cite this article in press as: Oreskovic, P., et al., Computer-generated vegan menus: The importance of food composition database
choice. J. Food Compos. Anal. (2014), http://dx.doi.org/10.1016/j.jfca.2014.07.002

589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622

You might also like