JOURNAL OF FOOD COMPOSITION AND ANALYSIS (2000) 13, 455}473doi:10.1006/jfca.2000.0890Available online at http://www.idealibrary.com on

ORIGINAL RESEARCH

An Application of Criteria to Evaluate Quality of Dietary FibreData in Brazilian Foods

Elizabete W. Menezes1, LuH cia Caruso, and Franco M. Lajolo

Departamento Alimentos e Nutrima8o Experimental, Fac. CieLncias Farmace(uticas, Universidade SaJ o Paulo,Cx. Postal } 66083, CEP } 05315-970, SaJ o Paulo } SP, Brazil

Received July 13, 1999, and in revised form February 16, 2000

Food composition data are very relevant to any study of human nutrition. An importantconsideration is the data control quality, which ensures the reliability of the information. It is notenough to have just data but their quality has to be evaluated. The analytical results will beconsidered unreliable if there is not a clear and accurate program of quality control. The dietary"bre (DF) was chosen as a model with the purpose of applying the USDA criteria used forevaluation of selenium, copper and carotenoids. The reason for this choice was the lack of dataabout this nutrient, and because of the important role DF plays in the prevention of chronicdiseases. Dietary "bre data were graded through standardized questions and decision pathwaysdeveloped for the categories listed as follows: number of samples, sampling plan, samplehandling, analytical method and analytical quality control. From these categories a con"dencecode (CC) was established and Brazilian foods could be graded. The CC obtained for 180 foodswere 29% with codes A and B (considerable and reasonable con"dence) and 71% with codeC (reduced con"dence). However, it was not possible to establish the CC for 244 compiled foodsdue to the inadequate methodology applied to many of them. The quality evaluation of the DFdata allowed not only assessment of reliable actual data, but the identi"cation of priority foodsfor analysis as well. The evaluated DF data was then included in a Brazilian Food CompositionDatabase (http://www.fcf.usp.br/tabela)which allowed users to obtain acceptable information onthe intake of dietary "bre (DF) by the population.

( 2000 Academic Press

Key=ords: data quality evaluation; dietary "bre; database.

INTRODUCTION

Data about food composition are very relevant to any study of human nutrition andthe necessity of updating that information is of international concern. The Interna-tional Network of Food Systems (INFOODS) has focused e!orts to generate regionalfoods composition tables.

In Brazil, a food composition project was developed with the purpose of updatingthe data of food composition and improving the quality level of information. Anotheraim of this project was to elaborate a Brazilian food composition database, accordingto the INFOODS/LATINFOODS standards for Latin America (Lajolo andMenezes, 1998).

1To whom correspondence and reprint requests should be addressed. Fax: 55 11 8154410. E-mail:wenzelde@usp.br

0889}1575/00/040455#19 $35.00/0 ( 2000 Academic Press

456 MENEZES, CARUSO, AND LAJOLO

Former studies have shown a lack of suitable data concerning dietary "bre (Lajolo,1995; Lajolo and Menezes, 1997), in spite of the great importance of this componentin human nutrition (HernaH ndez et al., 1995; Baghurst et al., 1996; Walker, 1998).According to the American Dietetic Association (1993), the ingestion of dietary "breassociated with a balanced diet (rich in carbohydrates and poor in lipids) is importantto promote health and to lower the risk of degenerative diseases. The nutritionalrecommendations proposed to the Brazilian population (Vannucchi et al., 1990)suggested a minimum daily ingestion of 20 g of dietary "bre.

The main problem in quanti"cation of dietary "bre involves the analytical method.A lot of data from Brazilian laboratories were analysed using crude "bre or detergentneutral "bre methods. It was only in the late 1990s that the enzymatic}gravimetricand non-enzymatic}gravimetric methods started being adopted in several laborator-ies (Lajolo, 1995; Lajolo and Menezes, 1997). Furthermore, in order to obtain data tobe included in a food composition database, other aspects must be considered besidesthe adequate analytical method; if there is not a de"ned quality assurance program,the analytical results will be restricted (Green"eld and Southgate, 1992).

The evaluation criteria for data quality in a laboratory must be concerned withseveral possible errors. An error associated with sampling plan is established whenthere is no representativeness or the sample handling was not suitable, mainly in thehomogenization stage, which is essential to guarantee the repeatability. The analyticalmethodology requires adequate methods and reagents besides the control of accuracyand precision levels. These errors can be eliminated almost completely through goodlaboratory practice (GLP) (Green"eld and Southgate, 1992; Marmo, 1997).

Quality criteria for food composition data evaluation were proposed in 1980, by theNutrient Composition Laboratory, Agricultural Research Service and NutrientDatabase* Human Nutrition Information Service * USDA (Holden et al., 1987).

Studies were developed to evaluate data quality referring to certain nutrients infoods: selenium (Bigwood et al., 1987; Gebhardt and Holden, 1993; Holden et al., 1987;Schubert et al., 1987), copper (Lurie et al., 1989), carotenoids (Mangels et al., 1993) andfolates (Holden et al., 1997).

The objective of this paper was to establish the quality of data on dietary"bre available in Brazil and the elaboration of a dietary "bre composition table.The work was based on an adaptation of USDA methodology for analysis of dataquality of other nutrients, which resulted in a speci"c model for "bre data qualityassessment.

METHODS

Database Elaboration

The Brazilian database was built up in our laboratory through data compilationfrom publications, monographs and internal laboratory data which were furtherevaluated.

A form was produced for use in compilation of food composition data (Menezes etal., 1997). The "rst part contains information about the food identi"cation. It is basedon the INFOODS guidelines for describing foods (Truswell et al., 1991). Theapproach is a broad and multifaceted descriptive mechanism using a series ofdescriptors.

The second part contains information about the nutrients (average content, vari-ation, standard deviation, units, methodology, edible part, conversion factor forprotein, and others) (Green"eld and Southgate, 1992; Klensin et al., 1989; Burlingame,

EVALUATION OF DIETARY FIBRE IN BRAZILIAN FOODS 457

1998). The last part is related to the quality of the data, providing consistent andobjective information for evaluation of published data on nutrient composition.

Data Quality Evaluation

Analytical quality evaluation of dietary "bre data was based on studies carried out byUSDA with other nutrients. According to parameters and criteria for selenium andcarotenoids (Holden et al., 1987; Holden, 1997; Mangels et al., 1993), adaptations wereintroduced so as to be suitable when applying to "bre data.

The system involved "ve general categories for data evaluation: (a) number ofsamples; (b) sampling plan; (c) sampling handling procedures; (d) analytical method;and (e) analytical quality control. These categories represent the major determinantsof data quality and are essential for their inclusion in the database. Speci"c criteriawere developed for each category; ratings for each ranged from 0 (unacceptable) to3 (highly acceptable).

Next, a quality index (QI), an indicator of the overall data quality in a food froma single reference or study, was calculated. In general, the mean of the "ve ratings wasdesignated the quality index (QI). When the rating for analytical method was zero orwhen the three ratings were zero, the quality index for that value was set to zero. A QIof one or more indicated an acceptable value and was retained (Holden et al., 1987;Holden, 1997).

Initially, similar foods were grouped into preliminary aggregates. For similar kindsof a single food, all acceptable data were grouped together under a general fooddescription.

For this general food description each QI was added and the con"dence code (CC)was obtained. The sum of quality indices, CC and meaning of CC were: '6, a, theuser can have considerable con"dence in this value; 3.4}6.0, b, the user can haveconsiderable con"dence in this value; but some problems exist about the data onwhich the value was based; 1.0 to (3.4, c, the user can have less con"dence in thisvalue because of the limited quantity and/or quality of data (Holden et al., 1987;Holden, 1997).

RESULTS AND DISCUSSION

The system used here re#ects the basic concepts described by Holden et al. (1987).Holden (1997) and Mangels et al. (1993). However, in order to make it possible toapply the quality criteria to Brazilian dietary "bre data, an adaptation of these criteriawas made and is described in Table 1. As can be seen, the quality evaluation rankedamong the number of samples, sampling plan, sample handling procedures, analyticalmethod and the analytical quality control.

In order to facilitate the application of these adapted quality control criteria,standardized questions and decision pathways (pattern schemes) were elaborated. Thepattern schemes below (1}5) refer to the score each datum can obtain in each category.

Pattern Schemes

Number of samples (Scheme 1). According to Holden et al. (1987), a precise statisticis necessary for the adequate number of samples to be studied. For this, the nutrientand the food must be considered. It is also important to indicate the standard

SCHEME 1. No. of samples.

458 MENEZES, CARUSO, AND LAJOLO

deviation (S.D.) or standard error (S.E.). In this category, score zero is not applied. Thecategory ranges from 1 to 3.

Sampling plan (Scheme 2). The analyst should be attentive not only to the analyti-cal method development but also to the sampling plan. The strategies which weredeveloped to select the food units as analytical samples are fundamental to guaranteethe representativeness (Holden, 1994).

SCHEME 2. Sampling plan.

TABLE 1

Quality criteria adapted for evaluation of dietary "bre data

Evaluation 3 2 1 0

No. of samples '10, standard deviation (S.D.)and/or standard error (S.E.)

'3 1}2, not speci"ed *

Sampling plan Sampling of multiple geographicalareas, with description ofstatistical basis for sampling,representative sample ofbrands/varieties consumed orcommercially used

Vegetal sources: from regionalcooperatives/emporiums.Industrialized products:#3 lots

Incomplete description orwithout any strategy developedfor selecting of foods

Not described

Sample handling Complete documentation ofprocedures, description ofmoisture control, homogenizationand characteristics of theportion food

Partial documentation ofprocedures

Limited documentation ofprocedures

Not described

Analytical method Complete documentation of themethod with validation,recoveries of 95/}/100% andcoe$cient of variation (cv.))5%

Enzymatic}gravimetric ornon-enzymatic}gravimetricmethods (for fruits withlittle starch), only informationof cv. (cv.)10%)

Enzymatic}gravimetric ornon-enzymatic}gravimetricmethods (for fruits withlittle starch), only informationof cv. (cv. 10}/15%)

Enzymatic}gravimetric ornon-enzymatic}gravimetricmethods (for fruits with littlestarch), cv. '15%

Analytical quality control Complete documentation andexcellent values, SRM mean$2 S.D. and cv.)5%

Complete documentation andacceptable values, SRM mean$21@2 S.D. cv. 5}/10%

Incomplete documentationand acceptable values, SRMmean$3 S.D. and cv. 10}/15%or only information of cv.(cv.)15%)

No documentation or noacceptable values

EV

AL

UA

TIO

NO

FD

IET

AR

YF

IBR

EIN

BR

AZ

ILIA

NF

OO

DS

459

460 MENEZES, CARUSO, AND LAJOLO

As the food universe is wide, a representative sampling plan elaborated from acareful statistical study is recommended (Green"eld and Southgate, 1992).

To evaluate the sampling plan, every information and variable must be considered.It is necessary to check the representativeness of the sample in relation to the availableand/or consumed food by the population concerned. Another point refers to thesample source. It is important to know if vegetable foods are from multiple geographi-cal areas and if the samples of industrialized products were collected in only one lot orin di!erent ones. Other further information is also necessary (e.g. the cultivar, productbrand).

The criteria established to the sampling plan (Scheme 2) are described below:

Score 3: When elaborated through strategies, which guarantee the representa-tiveness (statistical basis for sampling).

Score 2: Whenever the sample of the vegetable source was from a cooperative oremporium. When the products were industrialized, more than three lots were needed.

Score 1: Incomplete description or without any strategies developed.Score 0: Not described.

Sample handling (Scheme 3). Foods are unstable materials subject to attack bybiological agents (e.g. micro-organisms, enzymes). Heat, light, oxygen, moisture andcontaminants may change their chemical composition. In order to minimize nutrientlosses, some factors must be considered in sample handling (speed in this process, andnecessary care during transport and storage) (Green"eld and Southgate, 1992).

As moisture can interfere in the dietary "bre quanti"cation, it must be controlledduring the storage. Also there must be special care in the food package.

In the preparation of the analytical sample, it is necessary to have adequatelaboratory techniques and supervision in order to promote a suitable homogeniz-ation. This process will assure that the analytical portion is representative of thewhole.

SCHEME 3. Sample handling.

EVALUATION OF DIETARY FIBRE IN BRAZILIAN FOODS 461

The description of portion of the food analysed is essential. The ideal is to analysethe edible portion considering its variables (e.g. a fruit can be analysed peeled orunpeeled, a vegetable can be raw or cooked). The processing method (e.g. polished orintegral rice), cooking methods (kind and time) and the storage conditions (e.g. frozenfoods, fresh foods) must also be described.

So, the proposed scheme (Scheme 3) focuses on the detailed description of moisturecontrol, homogenization and the characteristics of the portion.

Analytical method (Scheme 4). The proposed pattern scheme considered only die-tary "bre methods usually used in Brazilian laboratories: enzymatic}gravimetric

SCHEME 4. Analytical method.

SCHEME 4. Continued.

462 MENEZES, CARUSO, AND LAJOLO

(Cho et al., 1997) or non-enzymatic}gravimetric (for foods with little starch) (Li andCardozo, 1992). It is necessary to take into consideration that the chemical methods(Wood et al., 1993; Cho et al., 1997), despite their world-wide importance andrecognition, are not used routinely in Brazil. They are used mainly to characterize thedi!erent components of dietary "bre fractions in speci"c products.

Besides choosing a suitable method it is important to consider the validation ofthe method in the laboratory. This validation consists of con"rming the resultsreliability of an analytical method application. This can be made through interlabora-tory studies, or through recuperation of standard reference material (SRM) or internalstandard. The percentage of recovery is calculated by the formula:

% recoveries"mean value]100/&&true'' value.

EVALUATION OF DIETARY FIBRE IN BRAZILIAN FOODS 463

The collaborative study held with the participation of nine Latin American countries(19 laboratories) and with the support of CYTED (Programa Iberoamericano deCiencia y Tecnologia para el Desarrollo) attested that nearly 37% of the participantshave not attained the acceptable results at the statistically de"ned control limits. Someprocedures used in di!erent laboratories could partly explain the attained results.Many laboratories showed problems related to the quanti"cation of protein and ashin the residues. Others have not carried out enzyme purity control or used thestandard reference material or similar (Filisetti, 1997).

Analytical quality control (Scheme 5). In this category, the accuracy and precisionof the method are considered for the nutrient analysed. Accuracy is the level atwhich the analysed value estimates the &&true'' value. The precision refers to theattainment of close values with minimum variation among themselves, and this canbe estimated through the coe$cient of variation (% cv."S.D.]100/mean). Thesmallest values of the % cv. represent the most precise analysis (Holden et al., 1987;Smith, 1994).

The accuracy determination depends on the quality control for each analysis. AnSRM, a secondary or an internal standard can be applied, in order to verify theaccuracy of the method before each analysis. When an SRM is used, the valuesattained should fall within the reference mean value$S.D. In relation to the secondarystandard there must be an intra or interlaboratorial study to establish the mean andS.D. It should be observed that when the accuracy and precision values di!er, thelowest value will determine the score of the analytical quality control (Holden et al.,1987).

SCHEME 5. Analytical quality control.

464 MENEZES, CARUSO, AND LAJOLO

The precision and accuracy values used to evaluate the analytical quality control(Scheme 5) are described below:

Score 3: Complete documentation and excellent valuesAccuracy * SRM results: mean$2 standard deviations (S.D.)Precision * % cv.)5%;

Score 2: Complete documentation and acceptable valuesAccuracy * SRM: mean$21

2S.D.

Precision * % cv.: 5}/10%;

Score 1: Incomplete documentation and acceptable valuesAccuracy * SRM: mean$3 S.D.

Precision * % cv.: 10}/15%;Or only precision documentation

Precision * % cv.:)15%;

Score 0: No documentation or no acceptable values.

Results Considerations

About 424 vegetal foods were listed as a result of the dietary "bre compilation data.The analytical method criteria showed the following distribution: enzymatic}gravimetric methods (33%), neutral detergent methods (32%), crude "bre method(13%) and non-analysed "bre (22%).

Many of the national data presented the crude "bre or detergent neutral fraction.Though these methods were described in AOAC, it was shown that they did notexpress the real value of the dietary "bre component (HernaH ndez et al., 1995;Baghurst et al., 1996; Cho et al., 1997). These data scored zero in the analytical methodcategory, which meant a quality index (QI) of nil. So it was impossible to establishtheir CC.

The 180 remaining data of the analyses carried out through enzymatic}gravimetricmethods (Cho et al., 1997) and the method proposed by Li and Cardozo (1992) werejoined and the quality criteria were applied for CC determination.

Table 2 provides the quality evaluation of dietary "bre data in broccoli and bread.These foods were selected to illustrate some problems found in the data evaluationand also how the CC was established.

The analysis of raw broccoli was present in two references. However, data from oneof the studies was considered unacceptable due to the 0 rating given to the analyticalmethod (considered inadequate) and the non-description of the sampling plan andanalytical quality control. The CC C based only on one QI (1.4) indicated that moredata of raw broccoli are needed.

Bread samples (4) were analysed with enzymatic}gravimetric methods. The laborat-ory controlled the activity and purity of enzymes, the analytical control was partialand not su$ciently described. The CC based on the sum of QI (7.2) indicated that theuser can have considerable con"dence in this value.

In Table 3, the "nal result of quality evaluation of dietary "bre data was shownaccording to the "ve quality criteria adopted. It shows the mean value for total dietary"bre (minimum and maximum values), the CC and the reference of the informationabout the foods usually consumed by the Brazilian population. For similar foods, theQI of each study was added, and the CC was determined as the result of the QI sumsfor this food.

TABLE 2

Quality evaluation of dietary "bre data in broccoli and bread

Quality criteria ratingsTotal dietary

Description No. of samples No. of samples Analyticalmethod

Samplehandling

Sample plan Analyticalquality control

Coe$cient ofvariation % cv.

Quality index(QI)

"bre (g/100 g)Mean

BroccoliRaw 1 1 0 1 0 0 01 1.88Raw 3 2 2 2 1 1 0.75 1.42 2.17

1.4 2.17Con"dence code (CC)"c

Bread, wheatWhole 3 2 2 2 2 1 3.57 1.82 3.92Whole 3 2 2 2 2 1 1.99 1.82 5.03Whole 3 2 2 2 2 1 6.71 1.82 4.32Whole 3 2 2 2 2 1 3.46 1.82 4.63

7.2 4.48Con"dence code (CC)"a

1The quality index is zero because of zero rating for analytical method.2Index '1 is required for inclusion of an individual dietary "bre value in the calculation of the mean.

EV

AL

UA

TIO

NO

FD

IET

AR

YF

IBR

EIN

BR

AZ

ILIA

NF

OO

DS

465

TABLE 3

Mean of total dietary "bre (TDF) and con"dence codes (CC) of Brazilian foods

TDF (g/100 g)Moisture

Food References (g) Mean Minimum Maximum CC

CerealsBread, gluten, loaf DANE (personal communication) 38.36 2.75 * * CBread, oat, loaf, diet DANE (personal communication) 34.69 3.58 * * CBread, rye, loaf DANE (personal communication) 34.40 5.03 4.82 5.24 BBread, wheat, black, loaf DANE (personal communication) 36.85 5.56 4.82 6.30 CBread, wheat, French DANE (personal communication) 24.50 3.00 * * CBread, wheat, hamburger DANE (personal communication) 31.03 3.14 * * CBread, wheat, loaf DANE (personal communication) 35.50 2.78 2.75 2.82 BBread, wheat, loaf DANE (personal communication) 29.31 2.95 * * CBread, wheat, loaf, diet DANE (personal communication) 37.71 3.69 2.85 4.53 BBread, wheat, loaf, light DANE (personal communication) 41.61 9.41 * * CBread, wheat, whole, loaf DANE (personal communication) 33.54 4.48 3.93 5.03 ABreakfast cereal, All Bran DANE (personal communication) 4.69 21.53 * * CBreakfast cereal, corn, Chococroc Lanfer Marquez et al. (1997) 4.24 2.82 * * CBreakfast cereal, corn, Corn Flakes Lanfer Marquez et al. (1997) 5.52 3.57 2.24 4.81 CBreakfast cereal, corn, honey, Kruska Lanfer Marquez et al. (1997) 5.70 2.72 * * CBreakfast cereal, corn, honey, peanut DANE (personal communication) 5.42 1.52 * * CBreakfast cereal, corn, sugar Lanfer Marquez et al. (1997) 4.05 1.93 0.93 3.11 BBreakfast cereal, Froot Loops Lanfer Marquez et al. (1997) 3.04 2.15 * * CBreakfast cereal, oat, nut, honey DANE (personal communication) 6.00 4.90 * * CBreakfast cereal, rice, Choco Krispis Lanfer Marquez et al. (1997) 4.00 2.64 * * CCorn, bran, Zea mays DANE (personal communication) * 39.78 * * CCorn, green, canned, Zea mays DANE (personal communication) 78.80 3.40 * * CCorn, meal, cooked, Zea mays Filisetti-Cozzi and Lajolo (1991) 73.10 1.40 * * CCorn, meal, raw, Zea mays DANE (personal communication) 10.52 3.68 * * CCorn, snack, Zea mays DANE (personal communication) 8.26 15.96 * * COat bran, Avena sativa DANE (personal communication),

Callegaro (1995) 10.85 14.89 13.80 15.98 COat, #akes, Avena sativa DANE (personal communication),

Filisetti-Cozzi and Lajolo (1991) 9.70 9.42 9.38 9.50 AOat, meal, Avena sativa DANE (personal communication) 9.80 10.26 * * CRice, bran, Oryza sativa DANE (personal communication) 4.98 24.34 * * CRice, white, cooked, Oryza sativa Pedrosa et al. (1994),

Filisetti-Cozzi and Lajolo (1991) 73.35 0.92 0.89 0.97 C

466M

EN

EZ

ES,C

AR

USO

,AN

DL

AJO

LO

TABLE 3 Continued

TDF (g/100 g)Moisture

Food References (g) Mean Minimum Maximum CC

Rice, white, raw, Oryza sativa Callegaro (1995) 12.90 1.65 * * CRice, whole, raw, Oryza sativa Callegaro (1995) 12.85 3.00 * * CRye, whole, meal, Secale cereale DANE (personal communication) 7.90 21.20 * * CSpaghetti, cooked Filisetti-Cozzi and Lajolo (1991) 71.50 1.30 * * CWheat, bran, ¹riticum vulgare Filisetti-Cozzi and Lajolo (1991) 11.00 45.00 * * CWheat, meal, ¹riticum vulgare DANE (personal communication) 2.85 * * C

<egetablesBean, cooked, Phaseolus vulgaris L. Pedrosa et al. (1994),

Filisetti-Cozzi and Lajolo (1991) 73.77 6.13 5.60 7.20 BBean, raw, Phaseolus vulgaris L. DANE (personal communication) 12.15 21.58 * * CBeet, cooked, Beta vulgaris DANE (personal communication) 92.85 2.24 * * CBeet, leaf, cooked, Beta vulgaris Almeida-Muradian et al. (1997a) 94.27 1.67 1.57 1.77 CBeet, leaf, raw, Beta vulgaris Almeida-Muradian et al. (1997a) 90.69 2.70 * * CCarrot, leaf, raw, Daucus carota Almeida-Muradian et al. (1997b) 81.62 7.91 6.90 8.91 CBroccoli, cooked, Brassica oleracea L. DANE (personal communication) 92.19 2.86 * * CBroccoli, raw, Brassica oleracea L. DANE (personal communication) 94.10 2.17 * * CCabbage, raw, Brassica oleracea L. DANE (personal communication) 96.02 1.37 1.34 1.39 CCarrot, leaf, cooked, Daucus carota Almeida-Muradian et al. (1997b) 92.13 3.53 3.19 3.95 BCarrot, raw, Daucus carota DANE (personal communication) 96.40 1.54 * * CCauli#ower, raw, Brassica oleracea L. DANE (personal communication) 93.86 2.05 * * CChayote, cooked, Sechium edule DANE (personal communication) 92.27 2.10 * * CChayote, raw, Sechium edule DANE (personal communication) 95.10 1.33 * * CChickpea, meal, Cicer arietinum L. DANE (personal communication),

Poltronieri et al. (1997) 11.00 14.50 12.00 19.20 BChickpea, raw, Cicer arietinum L. Poltronieri et al. (1997) 10.80 20.72 * * CCucumber, raw, Cucumis sativus L. DANE (personal communication) 97.10 1.00 0.88 1.11 CEggplant, raw, Solanum melongena L. DANE (personal communication) 93.96 2.51 * * CGreen bean, cooked, Phaseolus vulgaris L. Almeida-Muradian and Fiorini (1996) 93.90 1.99 1.83 2.15 CGreen bean, raw, Phaseolus vulgaris L. Almeida-Muradian and Fiorini (1996) 91.30 2.78 2.54 3.02 CLettuce, ¸actuca sativa L. Filisetti-Cozzi and Lajolo (1991) 97.52 1.10 * * CManioc, cooked, Manihot esculenta C. DANE (personal communication),

Pedrosa et al. (1994) 72.10 1.80 * * CManioc, meal, raw, Manihot esculenta C. DANE (personal communication),

Filisetti-Cozzi and Lajolo (1991) 7.60 5.60 5.50 5.80 C

EV

AL

UA

TIO

NO

FD

IET

AR

YF

IBR

EIN

BR

AZ

ILIA

NF

OO

DS

467

TABLE 3 Continued

TDF (g/100 g)Moisture

Food References (g) Mean Minimum Maximum CC

Manioc, yellow, raw, Manihot esculenta C. DANE (personal communication) 71.30 1.94 * * CMushroom, shiitake, dried, cru DANE (personal communication) 6.69 43.54 * * CMushroom, sun-dried, raw DANE (personal communication) 5.91 17.77 * * COkra, raw, Hibiscus esculentus DANE (personal communication) 91.00 3.60 * * COnion, raw, Allium cepa L. Filisetti-Cozzi and Lajolo (1991) 87.80 1.90 * * CPepper, red DANE (personal communication) 7.13 7.40 * * CPepper, sweet, raw, Capsicum annuum L. DANE (personal communication) 94.60 0.62 * * CPotatoes, cooked, Solanum tuberosum L. Filisetti-Cozzi and Lajolo (1991) 85.40 1.60 * * CPotatoes, sweet, cooked, Ipomoea batatas DANE (personal communication),

Pedrosa et al. (1994) 61.00 3.90 * * CPotatoes, sweet, raw, Ipomoea batatas DANE (personal communication) 65.90 3.87 * * CSoya, meal, without fat, Glycine max Filisetti-Cozzi and Lajolo (1991) 10.20 20.60 * * CSpinach, cooked, Spinacea oleracea DANE (personal communication) 94.76 1.60 * * CSquash, little, raw, Cucurbita verrucosa DANE (personal communication) 95.32 1.21 * * CSquash, raw, Cucurbita moschata DANE (personal communication),

Silva and Penteado (1990) 95.14 1.94 * * CSwiss, chard, cooked, Beta vulgaris DANE (personal communication) 95.80 1.25 * * CTomato, raw, ¸ycopersicum esculentum DANE (personal communication),

Filisetti-Cozzi and Lajolo (1991) 94.45 1.23 1.00 1.46 CTurnip, raw, Brassica napus DANE (personal communication) 96.30 1.11 * * CYam, cooked, Dioscorea bulbifera L. DANE (personal communication) 71.16 2.58 * * CYam, raw DANE (personal communication) 81.45 1.52 * * CYam, raw, Dioscorea bulbifera L. DANE (personal communication) 73.80 2.34 * * C

FruitsAcerola, frozen, Malpighia punicifolia Salgado et al. (1997) 92.00 1.85 * * CAcerola, Malpighia punicifolia Salgado et al. (1997) 92.00 2.49 * * CApple, Malus sylvestris Filisetti-Cozzi and Lajolo (1991) 85.20 2.00 * * CBanana, nanica, Musa spp. DANE (personal communication) 77.30 1.50 * * CCashew, Anacardium occidentale Salgado et al. (1997) 87.80 3.17 * * CCashew, frozen, Anacardium occidentale Salgado et al. (1997) 87.80 2.30 * * CGrape, frozen, <itis vinifera L. Salgado et al. (1997) 87.46 1.12 * * CGrape, <itis vinifera L. Salgado et al. (1997) 87.46 2.27 * * CGraviola, Annona muricata L. Salgado et al. (1997) 87.12 4.31 * * C

468M

EN

EZ

ES,C

AR

USO

,AN

DL

AJO

LO

TABLE 3 Continued

TDF (g/100 g)Moisture

Food References (g) Mean Minimum Maximum CC

Graviola, frozen, Annona muricata L. Salgado et al. (1997) 87.12 2.47 * * CGuava, red, Psidium guajava DANE (personal communication) 85.81 4.95 * * CGuava, white, Psidium guajava DANE (personal communication) 86.07 5.13 * * CJabuticaba, Myrciaria cauli-ora DANE (personal communication) 87.85 2.06 * * CMango, frozen, Mangifera indica Salgado et al. (1997) 82.11 2.58 * * CMango, Mangifera indica Salgado et al. (1997) 82.11 3.28 * * COrange, Citrus aurantium Filisetti-Cozzi and Lajolo (1991) 91.00 1.80 * * CSapoti, frozen, Manilkara zapota Salgado et al. (1997) 84.13 6.72 * * CSapoti, Manilkara zapota Salgado et al. (1997) 84.13 9.98 * * CSugar apple, Annona squamosa L. Salgado et al. (1997) 79.80 5.62 * * CSugar apple, frozen, Annona squamosa L. Salgado et al. (1997) 79.80 4.16 * * CSurinam cherry, Eugenia pitanga DANE (personal communication) 91.21 2.10 1.37 2.82 CUvaia, Eugenia uvalha DANE (personal communication) 85.53 2.04 * * C

Others (typical foods)00Baia8 o de dois'' DANE (personal communication),

Pedrosa et al. (1994) 67.60 2.90 * * C&&Cuscuz'' DANE (personal communication),

Pedrosa et al. (1994) 42.60 2.73 * * C&&Quibe'', fried DANE (personal communication),

Pedrosa et al. (1994) 56.90 2.00 * * CBean, &&tutu'' DANE (personal communication),

Pedrosa et al. (1994) 55.50 5.00 * * CPolenta, cooked DANE (personal communication) 73.10 1.40 * * CSoap, dried, with "bre DANE (personal communication) 7.44 25.17 * * CSquash, pureH e DANE (personal communication),

Pedrosa et al. (1994) 86.40 1.70 * * C

EV

AL

UA

TIO

NO

FD

IET

AR

YF

IBR

EIN

BR

AZ

ILIA

NF

OO

DS

469

470 MENEZES, CARUSO, AND LAJOLO

The results achieved in the quality evaluation data of Brazilian foods were 29%under codes A and B. Foods that were awarded with code A credibility were those inwhich the user can have considerable con"dence according to the value presented.Code B implied that the user can have con"dence in the value, but there were someproblems regarding the data on which the value was based. The other foods (71%)were classi"ed under code C, meaning that the user may have less con"dence in itbecause of the limited quantity and/or quality of data.

It is important to consider that codes A, B, or C are acceptable for inclusion in thedatabase (Mangels et al., 1993). On the other hand, the information with QI nil wasnot included, as it was considered unacceptable.

In a study where Mangels et al. (1993) evaluated the analytical quality data forcarotenoids in fruits and vegetables, 9% of foods received CC A. These results aresimilar to the ones attained in our quality evaluation of dietary "bre data (8% code A).

On the other hand, Lurie et al. (1989) evaluated data for copper and attained 14%of foods with CC A and Schubert et al. (1987) did the same with selenium, attaining25% of the foods with code A.

It is important to consider some di!erences when comparing results in di!erentstudies. Both copper and selenium are oligoelements appearing in small concentrationin foods (lg), and this increases the sharpness necessary in the analytical process forits detection. So the methodologies become more sensitive, with a higher level ofquality control, and that is one of the factors responsible for the higher percentage ofCC A in the analyses.

Lurie et al. (1989) found 62% of code C for copper because of the reduced numberof samples analysed (280). The result is similar to that obtained in the evaluation ofdietary "bre (71% showed CC C).

The high number of C graded foods was in part due to the lack of detailedinformation about the analytical procedures, insu$cient number of samples or poordescription of the analytical quality control used. Another factor was related to thenon-usage of SRM. Holden (1997) found it di$cult to achieve the highest score 3 whenassessing quality of USDA data of food composition. For instance, SRM, one of thepre-requisites for a 3 rating, was not available for many components. Lee and Prosky(1995) carried out an international survey of the use of SRM for dietary "bre analysis.The results showed that only 20% out of 147 laboratories used SRM. However, theuse of reference material is recommended and within the next years, more material isexpected to be available for analysts (Finglas, 1997).

Thus, the dietary "bre data were rigorously evaluated and then introducedinto the Brazilian Food Composition Database * USP (www.fcf.usp.br/tabela)(Lajolo and Menezes, 1998). The assessment also helped to de"ne priority foods forfuture analyses. Based on the information about the food most frequentlyconsumed by the Brazilian population, it could be observed that from the 43 mostconsumed vegetal foods, 10 were not analysed and 33 received code C and 2 codeB (Caruso, 1998).

The evaluated DF data allowed users to obtain acceptable information on theintake of DF by the population (Lajolo and Menezes, in press). A marked reduction ofDF intake has been evidenced during the last three decades. During the 1970s theaverage DF intake was 19.3 g/day, dropping to 16.0 g/day during the 1980s andhitting 12.4 g/day in the 1990s. This was due to the changes in eating habits, lifestylesand the socio-economic pro"le of the country.

The quality evaluation of dietary "bre data not only provided rigorously evaluatedinformation for professionals but also emphasized the need of awareness amongresearchers about the importance of adequate analytical procedures and detaileddescription. It also identi"ed the foods which are priorities for future analyses.

EVALUATION OF DIETARY FIBRE IN BRAZILIAN FOODS 471

It is also important to stress the need to better de"ne &&"bre'' and to develop newanalytical methods to characterize the "bre fraction and to relate it to functionalproperties (physical and biological). To be able to associate analytical data withphysical and functional properties ultimate goals should be de"ned. This is especiallyimportant considering the recent trend in nutritional labelling and functional fooddevelopment.

CONCLUSIONS

From the application of criteria to evaluate the quality of dietary "bre data inBrazilian foods, it was possible to identify several "bre analysis interfering factorswhich will help to obtain more sound data in the future. A CC of the data wasestablished and a revised dietary "bre table was, then, compiled and made available.Also as a result, priorities for future analyses could be established.

The authors acknowledge with gratitude the "nancial support granted by the Programa Iberoamericano deCiencia y TecnologmHa para el Desarrollo (CYTED), Projecto XI.6 &&ObtencioH n y CaracterizacioH n de FibraDieteH tica para su AplicacioH n en Alimentos para Regimenes Especiales'' and Conselho Nacional deDesenvolvimento CientmH "co e TecnoloH gico (CNPq).

REFERENCES

Almeida-Muradian, L. B. and Fiorini, F. (1996). Composiia8 o e valor proH -vitammHnico A das vagensmacarra8 o e manteiga, comercializadas em Sa8 o Paulo, Brasil. SEMINA Rev. Cult. Cient. ;niv. ¸ondrina17(1), 49}52.

Almeida-Muradian, L. B., Magri, L. M. and Rios, M. D. G. (1997a). Chemical characterization of Brazilianbeet leaves: provitamin A value with and without the saponi"cation step for raw and cooked samples.Bol. Chim. Farmaceutico 136(7), 537}542.

Almeida-Muradian, L. B., Popp, V. and Farias, M. P. (1997b) Provitamin A activity of Brazilian carrots:leaves and roots, raw and cooked and their chemical composition. CieL nc. ¹ecnol. Alim. 17(2), 120}124.

American Dietetic Association (1993). Position of the American Dietetic Association: health implications ofdietary "ber. J. Am Diet. Assoc. 93, 1446}1447.

Baghurst, P. A., Baghurst, K. I. and Record, S. J. (1996). Dietary "bre, non-starch polysaccharides andresistant starch. A review. Australia, CSIRO. Food Aust. 48 (Suppl.), 1}35.

Bigwood, D. W., Heller, S. R., Wolf, W. R., Schubert, A. and Holden, J. M. (1987). SELEX: an expert systemfor evaluating published data on selenium in foods. Anal. Chim. Acta 200, 411}419.

Burlingame, B. (1998). Proximate methods and modes of expression: variability as a harmonization issue.National Nutrient 22nd Databank Conference, Baton Rouge, Louisiana, June 20}22, 1996. http://www.nal.usda.gov/fnic/foodcomp/com/ [October 06, 1998].

Callegaro, M. G. (1995). Efeito do arroz integral em relaia8 o ao polido na mistura arroz- feija8 o sob algunspara( metros de avaliaia8 o nutricional em ratos em crescimento. Dissertaia8 o de mestrado, Faculdade deCie( ncias Farmace( uticas, Universidade de Sa8 o Paulo.

Caruso, L. (1998). Avaliaia8 o da qualidade analmH tica dos dados sobre "bra alimentar * um modelo.Dissertaia8 o de mestrado, Faculdade de Cie( ncias Farmace( uticas, Universidade de Sa8 o Paulo.

Cho, S., Devries, J. W. and Prosky, L. (1997). Dietary Fiber Analysis and Applications. AOAC International,USA.

Departamento de Alimentos e Nutriia8 o Experimental (DANE) Faculdade de Cie( ncias Farmace( uticas, USP(analyses elaborated in several laboratories of the Department, personal communication).

Filisetti, T. M. C. C. (1997). Estudo colaborativo para anaH lise de "bra alimentar. Bol. Soc. Bras. CieL nc.¹ecnol. Aliment. 31, 112}113.

472 MENEZES, CARUSO, AND LAJOLO

Filisetti-Cozzi, T. M. C. C. and Lajolo, F. M. (1991). Fibra alimentar insoluH vel, soluH vel e total em alimentosbrasileiros. Rev. Farm. BioqumHm. ;niv. S. Paulo 27, 83}99.

Finglas, P. M. (1997). Progress in analytical quality control for food composition work: assuring quality ofanalytical data. In 16th International Congress of Nutrition, pp. 310}311. NRC, Montreal, Canada.

Gebhardt, S. E. and Holden, J. M. (1993). Provisional ¹able on the Selenium Content of Foods.USDA/Human Nutrition Information Service [HNIS/PT-109], Rockville.

Green"eld, H. and Southgate, D. A. T. (1992). Food and Composition Data, Production, Management and;se. Chapman & Hall, London.

HernaH ndez, T., HernaH ndez, A. and Martinez, C. (1995). Fibra alimentaria: concepto, propriedadesy methods de analisis. Alimentaria 261, 19}30.

Holden, J. M. (1994). Sampling strategies to assure representative values in food composition data. FoodNutr. Agric. 12, 12}20.

Holden, J. M. (1997). Assessment of the quality of data in nutritional databases. Bol. Soc. Bras. CieL nc.¹echnol. Aliment. 31, 105}108.

Holden, J. M., Beecher, G. R., Doherty, R., Davis, C. and Finglas, P. M. (1997). Evaluation system for theassessment of the quality of folate composition data of foods. In 16th International Congress of Nutrition.NRC, Montreal, Canada.

Holden, J. M., Schubert, A., Wolf, W. R. and Beecher, G. R. (1987). A system for evaluating the qualityof published nutrient data: selenium, a test case. In Food Composition Data: A ;ser1s Perspective(W. M. Rand, C. T. Windham, B. M. Wyse, V. R. Young, and E. Tokyo, Eds.), pp. 177}193. UnitedNations University, Japan.

Klensin, J. C., Feskanich, V. L., Truswell, A. S. and Southgate, D. A. T. (1989). Identi,cation of FoodComponents for INFOODS Data Interchange. UNU, Tokyo.

Lajolo, F. M. (1995). Grupo de trabalho: composiia8 o de alimentos. Bol. Soc. Bras. CieL nc. ¹echnol. Aliment.29, 57}69.

Lajolo, F. M. and Menezes, E. W. (1997). Grupo de trabalho: composiia8 o de alimentos. Uma anaH liseretrospectiva e contextualizaia8 o da questa8 o. Bol. Soc. Bras. CieL nc. ¹echnol. Aliment. 31, 90}92.

Lajolo, F. M. and Menezes, E. W. (1998). ¹abela Brasileira de ComposimaJ o de Alimentos-;SP. Departmentode Alimentos e Nutriia8 o Experimental, Faculdade de Cie( ncias Farmace( uticas, Universidade de Sa8 oPaulo (USP), availablel: http://www.usp.br/fcf/tabela [1/10/1998].

Lajolo, F. M. and Menezes, E. W. Dietary "ber and resistant starch intake in Brazil: recommendations andactual consumption patterns. In Handbook of Dietary Fiber and Functional Foods (S. S. Cho, and M.Dreher, Eds.). Marcel Dekker, New York (in press).

Lanfer Marquez, U. M., Nishi, L. E., Barros, R. M. C., Filisetti Cozzi, T. M. C. C. and Penteado, M. V. C.(1997). Estudo da composiia8 o qumHmica de #ocos de cereais com e( nfase nos teores de fenilalanina. CieL nc.¹ecnol. Aliment. 17, 314}319.

Lee, S. C. and Prosky, L. (1995). International survey on dietary "ber: de"nition, analysis, and referencematerials. J. AOAC Int. 78, 22}36.

Li, B. W. and Cardozo, M. S. (1992). Nonenzymatic: gravimetric determination of total dietary "ber in fruitsand vegetables. J. AOAC Int. 75, 372}374.

Lurie, D. G., Holden, J. M., Schubert, A., Wolf, W. R. and Miller-Ihli, N. J. (1989). The copper content offoods based on a critical evaluation of published analytical data. J. Food. Compos. Anal. 2, 298}316.

Mangels, A. R., Holden, J. M., Beecher, G. R., Forman, M. R. and Lanza, E. (1993). Carotenoids content offruits and vegetables: an evaluation of analytic data. J. Am. Diet. Assoc. 93, 284}296.

Marmo, O. L. (1997). Qualidade e con"abilidade dos dados analmH ticos. In CieL ncia de Alimentos Avanmose Perspectivas na AmeH rica ¸atina (D.B. Rodriguez-Amaya, and G. M. Pastore, Eds.), pp. 52}54.UNICAMP, Campinas.

Menezes, E. W., Caruso, L. and Lajolo, F. M. (1997). Uniformizaia8 o internacional de dodos brasileiros decomposiia8 o de alimentos. Bol. Soc. Bras. CieL nc. ¹echnol. Aliment. 31, 93}104.

Poltronieri, F., Are( as, J. A. G. and Colli, C. (1997). Extrusion cooling and iron bioavailability of chickpea(Cicer arietinum L.) IAC * Marrocos. In Bioavailability 197 =ageningen. Wageningen.

Pedrosa, L. F. C., ArauH jo, M. O. D., Lima, E. B., Melo, M. S. O. N. and Gadeiro, L. M. T. (1994). AnaH liseQumHmica de preparaio8 es usuais em cardaH pios populares brasileiros. Rev. Nutr. P;CCAMP. 7, 48}60.

Salgado, S. M., Guerra, N. B., Filho, A. B. M. and ArauH jo, A. A. R. (1997). Tropical fruits frozen pulps:e!ect of the process on the "ber content. In 16th International Congress of Nutrition. NRC, Montreal,Canada.

Schubert, A., Holden, J. M. and Wolf, W. R. (1987). Selenium content of a core group of foods based ona critical evaluation of published analytical data. J. Am. Diet. Assoc. 87, 285}299.

EVALUATION OF DIETARY FIBRE IN BRAZILIAN FOODS 473

Silva, M. E. M. P. and Penteado, M. V. C. (1990). Teores de vitamina C em alimentos de consumo habitualna regia8 o de Sa8 o Paulo. Dissertaia8 o de mestrado, Faculdade de Cie( ncias Farmace( uticas, Universidadede Sa8 o Paulo.

Smith, J. S. (1994). Evaluation of analytical data. In Introduction to the Chemical Analysis of Food (S.S.Nielsen, Ed.), pp. 51}63. Jones and Bartlet, London.

Truswell, A. S., Bateson, D. J., Mada"glio, K. C., Pennington, J. A. T., Rand, W. M. and Klensin, J. C.(1991). INFOODS Guidelines for describing foods: a systematic approach to describing foods to facilitateinternational exchange of food composition data. J. Food Compos. Anal. 4, 18}38.

Vannucchi, H., Menezes, E. W., Campino, A. O. and Lajolo, F. M. (1990). AplicamaJ o das RecomendamoJ esNutricionais Adaptadas a PopulamaJ o Brasileira. Cadernos de Nutriia8 o. SBAN 2, Legis Suma, Ribeira8 oPreto.

Walker, A. R. P. (1998). Fibre in health/disease * what now? Nutr. Res. 18, 607}614.Wood, R., Englyst, H. N., Southgate, D. A. T. and Cummings, J. H. (1993). Determination of dietary "bre in

foods* collaborative trials. IV. Comparison of Enlglyst GLC and colorimetric measurement with theProsky procedure. J. Assoc. Publ. Anal. 29, 57}141.