Contents - Previous - Next

This is the old United Nations University website. Visit the new site at http://unu.edu

Weaning blend criteria

Guidelines for formulating both home-prepared and commercially processed weaning blends are given in table 1. The computer programme will adjust ingredient types and proportions within these guidelines to specify either the mixture having the minimum cost at a selected protein quality level (J. 1. Wadsworth, personal communication, 1992), or, for poorer protein quality commodity sets, the blend that most closely approaches the table I protein quality standard [1].

The values presented for home-prepared blends were principally derived from Protein Advisory Group of the United Nations guideline no. 8, concerned with protein-rich mixtures for use as weaning foods [29], with supplementation by the Food and Agriculture Organization (FAO)/World Health Organization (WHO) Codex Alimentarius Commission's Committee on Foods for Special Dietary Uses [30] and more recently modified [31]. Yeast is limited to 3% to control the purine content of the diet [1].

Home-prepared blends have been formulated to provide protein-energy in the range of 12% to 15% [31]. To provide adequate caloric density, the table 1 standard formula for this type blend specifies 11 g fat per 100 g dry weight, which results in energy derived from fat in the order of 20% to 25% [30]. The alternative formula provides for substitutions of sugar for some of the fat (2 g sugar for I g fat) [31]. The 8% level of sugar was based on field test experience [32].

For home-prepared blends, crude fibre was limited to 5 g per 100 g dry matter, in accordance with published recommendations [31]. For young children, the general consensus is to keep fibre intake low [8].

Among the various reasons are possible laxative effect produced by slow fermentation by intestinal flora, lower caloric density due to increased bulk, irritation of the gastrointestinal mucosa, reduced digestibility, and reduced availability of vitamins and minerals.

TABLE 1. Formulation guidelines for home-prepared and commercially processed weaning blends

a. Guidelines from refs. 29, 30, 31, and 33.
b. The absorbed protein level refers to the blood level. See text.
c. The absorbed amino acid score represents the amino acid score of absorbed amino acids in the blood. See text.
d. Guidelines from refs. 30, 31, 33, and 35.
e. g/100 g actual weight.

The chemical analysis of crude fibre is an underestimate of true poorly digestible dietary fibre, and dietary fibres are of different types. Also, data are insufficient on the influence of the intake level of fibre on the possible adverse effects cited. For these reasons, different authors and organizations have set different maximum limits for fibre in weaning foods. In the present study, the high crude fibre level of 5% was chosen to permit a wider variety of market ingredients to be used in home-prepared blends. The maximum 2% (dry weight basis), as required in the US Food-for-Peace Program type blends [13], was specified for commercially processed blends, which select from a set of ingredients generally lower in fibre.

A maximum moisture limit was not set for the homeprepared blends, since neither FAD/WHO [30] nor Hofvander and Underwood [31] set a moisture limit. It seems reasonable to assume that storage time would be less before consumption of home prepared blends than for commercially processed blends. For the latter, long shelf storage could result in appreciable chemical deterioration at higher moisture levels. Therefore, for these blends, a moisture limit of 10% (dry weight basis) was set, as required for US Food-for-Peace Program blends [13].

The absorbed protein level represents the total blend crude protein multiplied by the fraction overall blend digestibility. The overall blend digestibility is estimated by using digestibility values for individual ingredients [27, 33, 34] and calculating a weighted mean in accordance with WHO [33]. The individual ingredient digestibility values are used in the same weighted manner to determine an absorbed amino acid score for any given amino acid in the blend.

The term "absorbed" as applied to blend protein level or amino acid score refers to the blood level. The protein quality recommendation by FAD/WHO [30] for a minimum amino acid score of at least 65 was specified as applicable to the absorbed protein level for both home-prepared and commercially processed formulations.

For both types of weaning blends, the caloric value is calculated on the basis of available energy rather than apparent energy. The apparent energy is calculated by applying the Atwater factors (4 kcal/g for protein and carbohydrate, 9 kcal/g for fat) to the appropriate proximate analysis values. Available energy is determined by multiplying the apparent energy by a factor related to the fibre content of the diet [33]. In this study, the factors of 1.0, 0.975, and 0.95 were assigned to food mixtures with crude fibre levels below 1.0%, 1.0% to 2.0%, and above 2.0%, respectively.

The formulation guidelines for commercially processed blends are a modification of those for corn (maize)-soy blend [35]. An earlier study indicated that corn (maize)-soy blend has a true nitrogen digestibility of 83% [14]. The true digestibility for the reference protein (milk, egg, meat) has been set at 95% [33]. The absorbed protein level corresponding to the 16.7% minimum specification [35] would thus be 16.7 x 83/95 = 14.6% (dry weight basis). This was increased to the effective protein level of 15.0% for commercially processed blends to make it come parable to that for home-prepared blends. The minimum protein quality recommendation of FAD/WHO [30] for an amino acid score of at least 65 was adopted also for commercial blends as applicable to absorbed amino acids.

In the corn (maize)-soy blend formulation, fat was kept within the specified limit to control rancidity during shelf storage. Donated foods such as corn (maize)-soy blend often are prepared at a high protein level, with the understanding that the recipient countries would be able to make up the food calories with added commodities [36]. It was recommended that fat be added with or without sugar to such foods during the time of preparation for eating to provide adequate caloric density [31]. In accordance with this suggestion, at the time of preparation of the pore ridge, either the stated quantity of fat alone (standard addition) or the given quantities of fat plus sugar (alternative addition) were combined with the commercially processed formulations. Specific instructions for doing this should appear on packages containing the mix.

The amino acid score of a blend is the score of the most limiting essential amino acid. It is a measure of the percentage of adequacy of a given essential amino acid with respect to human nutritional need. It is defined as follows [33]:

Amino acid score

=

The FAO/WHO/UNU Expert Consultation, reporting in 1985 on energy and protein requirements, recommended that the amino acid patterns of human milk should be accepted as the requirement for infants [33]. This pattern would more than meet the requirement patterns of all age groups. Since the weaning interval extends from infancy to about two years of age and possibly longer in Lusaka [5], this pattern for infants was selected as the reference pattern for the present study. However, the pattern for children age two to five years has also been widely used for this purpose where breast-feeding is prolonged.

TABLE 2. Essential amino acid patterns for human infants^a,b and preschool children^b

a. Amino acid pattern of human milk.
b. From ref. 33.

Both essential amino acid reference patterns are shown in table 2. To calculate absorbed amino acid score, the following must be taken into account: the level of each essential amino acid in each commodity protein, the fraction of each commodity protein in the total protein, and the true nitrogen digestibility of each commodity protein with respect to the reference protein. The composite absorbed sum of a given amino acid from the various contributing ingredients can then be computed. The percentage of this value with respect to the content of the same amino acid in the reference protein is then determined.

Both weaning blend formulations were designed to meet the protein and energy needs of preschool children. However, it is also important to meet other nutritional needs. When vitamin and mineral supplemeets are not included in the blends, emphasis must be placed on instructing mothers that additional foods that supply vitamins and minerals should be part of the children's diet along with porridge. In Nepal, health workers teach mothers to add chopped leafy greens or other family pot vegetables to the pore ridge [7]. A multimix concept is illustrated diagrammatically by "the food square" to teach mothers how to provide in a practical way for adequate vitamin and mineral intake together with proper protein and energy nutrition [37].

A thorough nutritional analysis would have to be made of formulations suggested by this present study to find deficiencies and to determine what dietary additions, from locally available foods, are required to make up for these deficits. Computer optimization can be extended beyond the present application to select mixtures of acceptable ingredients that would most closely meet both protein-energy requirements and limiting micronutrient requirements.

Strategy for alternative formulations

A blended weaning food will not be offered unless it is acceptable to both the mother and the child. Organoleptic (taste, odour, texture, appearance), economic, and cultural acceptability are all important. In a study of weaning practices and foods in Lusaka [5], a frequency of preferred use index was calculated from mothers' responses for each of a series of different market foods, assessing how often any given item would be used in weaning.

The top-rated protein- and/or energy-providing commodities were designated mother-favoured, and special formulations were developed with only those items as ingredients. However, results from a dietary intake survey of children, conducted as part of the same study [5], showed that some foods (i.e., eggs, fresh milk) that ranked high on the mothers' preference list were not actually used very much in feeding. Mothers would be inclined to give such foods if their circumstances and priority commitments permitted them to do so.

Ingredient cost is known to be an important acceptability factor. For this reason, and because preferred ingredients may not always be available, formulations were developed using selections from all available market items that provide substantial contributions of protein and/or energy. In the case of commercially processed weaning mixtures, since the mother does not actually see what items constitute a final blend, the organoleptic qualities of this finished product are likely to be more important than the ingredients in determining acceptability.

Space limitation permits only a few illustrative examples of the various computer-derived options detailed in the tables. During the next phase of research, promising weaning mixtures from among the mother-favoured ingredients and those permitting choices from among all available market commodities are to be screened by Zambian mothers to determine comparative organoleptic acceptability [32]. Further screening by children of blends that represent unusual mixtures of foods will be necessary to ascertain acceptability and gastrointestinal tolerance [32].

A suitable level of protein quality has to be designated in formulating blended weaning foods. The FAD/WHO minimum recommended amino acid score for such mixtures of 65 [30] was adopted for this study, as applied to absorbed amino acids. The codex committee also recommended that the protein quality level be above their stated minimum. This will generally increase blend cost but possibly may enhance organoleptic acceptability.

Clinical trials showed that vegetable protein mixtures relatively near to, but less than, milk in biological value were suitable for rehabilitation of children over one year of age with kwashiorkor [38]. This was true provided that the mixtures were fed at an adequate protein intake and were not markedly inferior to milk in protein quality. On the condition that acceptability is satisfactory, an increased intake of protein can compensate to some extent for a slightly deficient amino acid pattern.

The results shown in the tables are reported for various mixtures at the following protein quality levels: the highest absorbed amino acid score attainable by computer design, approaching, but yet below, the minimum of 65, when purified amino acids are not used in commercial blends; the specified level of 65 or more when formulated also for minimum cost; and an arbitrarily chosen value intermediate between the minimum standard and the highest attainable by computer formulation. Formulation at an intermediate absorbed amino acid score sometimes results in a considerably different mix of ingredients than is found in a minimum-level blend. Hence, differences in organoleptic acceptability are likely. As noted, comparative organoleptic acceptability (and, if necessary, gastrointestinal tolerance testing) is projected for selected mixtures during the next phase of research.

Home-prepared blends

The following approach was followed for developing formulations for these mixtures:

1. All mother-favoured ingredients [5] were considered that were available for a given season and that used a designated source of concentrated energy (fat, or fat plus sugar).

a. Ingredients were selected and their proportions were determined to produce the lowest cost blend that still meets formulation guidelines. In some instances, blends formulated from an assigned commodity set to meet table 1 guidelines at least cost actually have an absorbed amino acid score greater than 65.
b. Ingredients were selected and their proportions were determined to produce the lowest-cost blend at a protein quality level halfway between the highest obtainable by computer composition and that which still meets formulation guidelines (absorbed amino acid score > 65).

2. All survey ingredients available for the given season were considered. The same protocol was adopted as in a and b.

3. The optimum formulations under 1 and 2 were modified to investigate the effect of potentially advantageous ingredient changes on overall formulation, cost, and calculated nutritional variables. Such modifications may be of interest because of likely improvement in organoleptic acceptability, decreased chance of gastrointestinal problems, or the convenience of being able to substitute a similar food ingredient that may happen to be available at a given time. Appropriate constraints were introduced into the programme to explore such possibilities.

Commercially processed blends

A similar general strategy was used in computing alternative formulations for these blends. For mixtures that incorporated purified amino acids, the least expensive compositions were determined for a minimum absorbed amino acid score and for a somewhat higher protein quality. The higher level arbitrarily chosen was one-third between the minimum standard amino acid score of 65 and the highest obtainable by computer formulation. In the case of commercially processed blends, for each set of constraints, formulations were prescribed both with and without added chemical supplements.

Mixtures without addition of purified essential amino acids were not able to meet the minimum absorbed amino acid score of 65. These blends were formulated by computer optimization to attain the highest possible score approaching the minimum. As indicated earlier, such formulations can still be nutritionally suitable [38]. Blend formulations for two conditions of chemical supplementation have been prescribed: essential amino acid fortification with and without vitamin-mineral-antioxidant addition.

Results and discussion

Commodity availability, processing loss, and costs

Table 3 shows the market costs of home-prepared blend ingredients, the reasons for and quantitative extent of processing loss, and the actual cost of ingredients in the form in which they would constitute a weaning blend. It also indicates the seasonal availability of the individual foods in Lusaka.

As shown in table 4, for commercially prepared blends, ingredient cost at the point of extruder processing represents an adjustment of original commodity market cost for both pre-extruder material loss and the expense of the processing operations. To conserve nutrients, any vitamins, antioxidants, minerals, or purified amino acids would be added after processing the other ingredients in the extruder. All of the potential ingredients for these blends are available in commercial quantities throughout the year.

Tests on home-prepared ingredients

Table 5 compares particle size profiles for two homeprepared ingredients with that for one of the maize meal products available in Lusaka for making pore ridge and nshima. It should be pointed out, however, that maize or rice would probably not be milled at home or in the village by mortar and pestle to the same degree of fineness as achieved with the Christy mill. The particle size distribution of the Christy millground, roasted maize meal, except for the percentage of material passing through the 0.25-mm screen, is quite similar to that for commercial maize meal. A difference in the degree of softness of maize and rice kernel endosperms can account for the difference in particle size distributions for these grains after grinding in the Christy mill.

TABLE 3. Economic and availability information for home-prepared weaning blend commodities

NA = not applicable, commodity not processed prior to blending; C=cleaning; RS= roasting; C=grinding; D = dehulling; Rcp = removal of congealed material after pounding; Rvp = removal of veins after pounding; Rfp = removal of fibrous matter after pounding; Rsbp = removal of scales, bones, etc., after pounding; Rs = removal of shell.
a. The market prices are for March to June 1989, except caterpillars for November 1989. The exchange rate for June 1989 was 12 Zambian kwacha (K) per United States dollar (US$).
b. Seasonal availability: 1-12 represents January-December.

Measurements of the degree of doneness for cereal gruels made from commodities roasted for different time periods are listed in table 6. A commercial maize meal is used for comparison. The results show that sorghum gruels had quite different viscosity from maize and rice gruels. As length of roast increased, sorghum gruels decreased in viscosity, which is desire able from the standpoint of caloric density. The vise cosity of cooked rice gruel peaked with time.

Viscosity can change through alterations in starch structure by various heat-processing and other ape preaches [39]. The purpose of the present study was to determine appropriate roasting times, rather than to investigate causes of viscosity effects. Prolonged roasting can result in a brown colour and, of course, an expected accompanying deterioration in protein quality. A 15-minute roast would seem reasonable from the standpoint of adequate gelatinization, as reflected by viscosity values and observations on texture and softness. Table 7 shows the results of the softness test on roasted oilseeds and non-oilseed legumes, and on some other potential weaning blend ingredients that were not roasted. It is apparent that in preparing some quick-cooking weaning gruels, certain ingredients may have to be presoaked or the gruels may have to be cooked longer than others. The results of the urease activity test on roasted soy samples indicated a pH increase of 2.32 units for soya beans roasted 10 minutes and an increase of 0.01 for soya beans roasted for 15 minutes. The 10-minute roast was not long enough for adequate urease inactivation [13]. The 15minute roast showed a pH change below the recommended minimum of 0.05 for soy flour used in instant corn (maize)soy milk [13]. This means that there was some degree of overcooking, with some additional loss in protein quality likely [18]. However, a 15-minute roast is as close as can be specified, since under actual home-preparation conditions variability in the frequency of stirring and intensity of heat from the burning fuel influence roasting time.

TABLE 4. Economic information for commercially processed weaning blend ingredients

NA = not applicable, commodity not processed prior to blending; C = cleaning; G = grinding; DH = dehulling; DR = drying; DS = deskining R = removal of undehulled kernels.

a. The wholesale commodity prices are for March to June 1989. All commodities listed are available in wholesale quantities throughout the year. The exchange rate for June 1989 was 12 Zambian kwacha (K) per United States dollar (US$).
b. The estimates of total processing cost for the indicated operations as percentage of the raw commodity price per kilogram were based on industrial experience in the United States.
c. The cost of amino acids includes shipment from Japan.
d. The combined cost of the vitamin/antioxidant premix and the mineral premix includes shipment from the United States. The compositions of both the vitamin/antioxidant premix and the mineral premix are given in ref.35.

Chemical analyses

Proximate and essential amino acid analyses, together with estimates of protein digestibility, were

TABLE 5. Sieve analysis comparisons for roasted maize flour roasted rice flour, and commercial maize meal performed for ingredients of both types of blends but are not given because they have to be determined for each new application.

a. Text and refs. 14 and 15 describe sieve analysis.
b. Text describes roasting procedure.
c. Ground through a ().3-mm screen in a Christy Mill (8" lab mill, 8.000 rpm).

Home-prepared formulations

Table 8 gives examples of alternative home-prepared formulations with calculated nutritional and cost values, considering mother-favoured ingredients, concentrated energy coming from both fat and sugar, ingredient availability at different times of the year, and variation in specified protein quality. Similar calculations were made for mixtures from mother-favoured ingredients using concentrated energy from fat only, and also for blends permitting selection from all available ingredients, alternatively for fat only or fat plus sugar to enhance caloric density.

Table 9 gives compositional and calculated changes that may occur in home-prepared blends if alterations are made in original computer-optimized formulations. For example, it may be desirable to include either fresh milk or non-fat dry milk in a weaning food mixture because of what is available in the household at the time. Eliminating fresh milk or nonfat dry milk will increase the cost of the mixture by about 18% and 72%, respectively. Both new formulations represent pronounced changes in proportions and types of ingredients found in the original mixture, although there is little difference in the absorbed amino acid score among the original and two new blends. On the other hand, substituting dried cowpea leaves for dried pumpkin leaves results in a new blend of quite similar ingredient proportions, with about the same cost and nutritional values as the original.

Commercially processed formulations

Table 10 shows commercially processed formulations with their respective calculated nutritional and cost values, considering mother-favoured ingredients with fat plus sugar as concentrated energy sources, added chemicals (none, amino acids only, and amino acids plus vitamin-antioxidant-mineral mix), and protein quality-cost variations. Similar calculations were made for mixtures from mother-favoured ingredients using concentrated energy from fat only, and also for blends permitting selection from all available ingredients, alternatively for fat only or fat plus sugar to enhance caloric density. The variations in nutritional value and cost of such mixtures will be country-specific and are not described in further detail.

The cost and compositional differences as a result of home-added fat only or home-added fat plus sugar as energy source(s) were evaluated. Within the individual categories of mother-favoured ingredients in Zambia, there is little difference in dry weight cost between blends with fat only and those with fat plus sugar. There is also little difference between them in absorbed amino acid score in corresponding constrains categories. Comparison of the fat-only versus fat-sugar variation among the blends containing mother-favoured ingredients shows that generally the same ingredients were called for, but that their proportions were somewhat different.

TABLE 6. Texture/softness,^a appearance, and viscosity^b of porridges made from unroasted and roasted cereal commodities considered for use as home-prepared blend ingredients

a. The text describes the technique used for estimating softness. Other visual observations concerning texture are included.
b. The method for determining viscosities on both uncooked and cooked porridge with a Brookfield model RVT viscometer is described in the text. .
c. The methods used for roasting and, in the case of sorghum, for dehulling are found in the text.
d. For this observation, porridges were cooked according to softness evaluation, described in the text.

Costs of home-prepared and commercially processed blends are not directly comparable, because allowances for cooker-extruder processing and for subsequent product handling and packaging, marketing, and profit mark-up are not included in the price per kilogram dry weight for commercial blends.

It is not always possible to modify optimum formulations to obtain desired alternatives that still meet the basic criteria of table 1. For example, a maize-bean-groundnut mixture having 41.9% as is of beans might cause flatulence. It would seem desirable to restrict the bean content to a lower percentage while retaining the absorbed chemical score at the same level of 74.2%. However, even lowering the as is bean percentage by a small decrement to 39.5% produced an unfeasible solution (i.e., absorbed protein content of 15% could not be attained). To reduce the bean content and retain the specified absorbed protein level of 15%, using allowable commodities, it is necessary to increase the fat restraint level from 7%. This is because beans are high in protein and low in fat, whereas maize and groundnuts are lower in protein and higher in fat. Thus, to maintain the specified protein level when the bean content is reduced, adding more maize and groundnuts also means adding more fat.

continue

Contents - Previous - Next