Cover Image
close this book Food Composition Data: A User's Perspective (1987)
close this folder International food composition data
close this folder Nutrient intake data calculated using food composition tables: factors affecting accuracy
View the document (introductory text)
View the document Introduction
View the document Materials and methods
View the document Results
View the document Discussion
View the document References



It should first be emphasized that our aim was not to construct a Food Composition Table for local foodstuffs but, rather, to evaluate the degree of accuracy that can be expected from nutrient-intake data calculated from a Food Composition Table for Use in Latin America (i.e. recommended for this purpose).

It is apparent that differences of up to one order of magnitude can be found between the results of our local analysis and the values for some nutrients given by the INCAP table (the vitamin C content of green pepper determined here was 20 times higher than that given by the table). Of particular interest was the observation that nearly all the values for iron content were below 50 per cent of the value given by the INCAP table. It can readily be appreciated that the use of the table leads to an overestimation of iron consumption, and to reconciling the dietary data with the high prevalence of iron-deficiency anaemia found in the region [2, 11,22]. It is also interesting to note that no value for the iron content of the 20 foodstuffs analysed here fell within 80-120 per cent of the value in the INCAP table. Equally interesting is the fact that the protein content of the so-called "sources of protein" showed little difference between the two values compared here. This might be the starting-point for suggesting that nutrient composition data could be divided into two categories: those of nutrients that show a high variation - probably attributable to regional differences (soil, climate, season, species) - and those of nutrients in some foodstuffs that show very little variation, probably insignificant for dietary evaluation purposes. Minerals and some vitamins are likely examples of the first category, while protein - being a compulsory component of foodstuffs derived from animal or plant tissues - could be a good example of the second category. Appropriate software for identifying the members of each category could be easily developed, and there are probably enough data available from various food composition tables to be used for this purpose.

Table 2. Nutrient composition of local foodstuffs, Recife

  INCAP table no. Moisture Protein Fat Carbohydrate Ash lron Vitamin A Vitamin C
Foodstuff   (%) (%) (%) (%) (%) (mg/100B) (µg/l00g) (mg/l00g)
Pork blood 549 77 17.53 0.12 3.56 1.79 14.70 - 1.31
Pork liver 552 56 26.98 2.30 13.32 1.40 1.40 4,441 40.88
Pork heart 540 68 20.37 3.92 9.58 2.73 14.00 - 5.06
Coriander 143 90 1.10 2.75 3.14 2.99 11.60 1,291 129.00
Green onions 142 89 1.20 3.55 4.63 0.82 0.31 268 47.81
Green pepper 80 92 0.24 3.33 3.60 0.53 0.22 270 190.50
Tomato 271 95 0.49 0.55 2.92 0.65 0.09 368 16.69
Onion 137 85 0.21 2.41 11.57 0.55 0.20 - 11.62
Corn flour 27 11 8.62 1.76 77.54 0.56 0 26  
Beef, lean 579 63 21.93 6.25 7.61 1.12 0.06 - 2.02
Pumpkin 127 68 0.71 0.14 26.45 2.29 0.27 1,526 16.88
Sweet potato 108 64 1.57 0.10 31.02 0.81 0.04 10 26.63
Sweet potato, yellow 107                
Potatoes 242 86 0.71 0.11 11.63 1.27 0.34 0 32.00
Plantain 438 58 1.16 0.59 43.46 1.25 0.30 88 34.03
Cabbage, wild 149 90 4.63 0.17 3.19 1.32 1.44 - 190.69
Cassava flour 276 7 0.52 0.18 88.68 1.14 0.98 - 13.50
Beans, mulatinho 481 13 25.52 1.78 52.38 3.54 4.97 - 2.26
Beef, dried 582 28 56.80 16.32 17.07 15.95 6.90 - 1.13
Bacon 682 7 0.27 80.04 11.05 1.64 - - 1.16
Anguria - 91 1.35 0.05 4.74 0.41 0.34 - 36.38
Okra 252 83 1.61 0.06 12.85 1.21 0.56 14 27.06

Table 3. Estimated iron and vitamin A intakes using nutrient composition figures from INCAP table and analysisa

Nutrient Table Analysis
Protein g/d 20.9 21.7
Iron, mg/d 10.5 3.0
Vitamin A, µg/d    
(as retinol equivalents) 143.2 174.9

a. Food consumption data are from a survey in Agua Preta, Pernambuco [2].

Vitamin A nutriture constitutes a problem that should be looked at - in our region - from another angle. It is apparent that the difference in the figures for consumption resulting from the use of the INCAP table and our results is not enough to explain a lack of prevalence of severe signs of vitamin A deficiency [2, 6, 7, 12] which is not compatible with the very low vitamin A intakes reported in several surveys [2, 9, 12]. One possible explanation might be that some regional fruits, with a very high carotene and carotenoids content, are consumed by the population but not reported in the surveys. We have observed that the fact that some of these fruits are not actually "bought" may lead the population not to consider them as "foods." Thus, a significant contribution to vitamin A intake may have been overlooked in the past.

The problem of regional differences in nutrient composition - and the difficulties generated by the use of food consumption tables which are, most of the time, inadequate for specific situations - is well known. Our data have only shown what the practical implication of this may be, and one way to reconcile dietary data with other indicators of the nutritional status. Our data on "dish-nutrient composition" (fig. 1) shows another very serious drawback in the analysis of survey data with the aid of food composition tables: the so-called "foods-as-eaten" problem. In theory, there has never been any reason to consider as reliable "recipe composition data," i.e. the compound nutrient composition of a dish, obtained by addition of the contribution of each single (raw) ingredient.

Fig. 1. Dish composition: calculated v. analysed values.

This approach ignores liability to heat of a great many nutrients, and losses that may result from chemical reactions as a consequence of the interaction between ingredients which are "incubated" for variable periods at 100°C or more. From a chemical view, feijoada was the dish to undergo the most drastic treatment (see recipe) and, in keeping with this, losses of 66 to 95 per cent were observed for protein and vitamin A respectively, regardless of the nutrient composition value used for the "recipe calculation." It was beyond the scope of this work to determine the actual causes for these losses, and we are first to admit that that of protein was the most intriguing. But the obvious conclusion is that food composition tables cannot continue to be used, without restriction, for calculating nutrient intakes.

The number of foodstuffs and nutrients analysed here was very modest. One should bear in mind, however, that sweet potatoes, cassava flour, sugar, beans, and a little meat account for more than 70 per cent of the daily food intake of the underprivileged in this region [2], as well as in most of the rest of the country. This is what we consider to be "alimentary monotony." Regarding nutrients, our contention is that emphasis should always be given to those capable of generating, by deficient or excessive intake, public health problems. These nutrients would include protein, vitamin A, and iron; the list would vary according to region, but would certainly be limited in each case.

It is becoming increasingly important to count on reliable sources of accurate data for nutrient intake evaluation in connection with a number of nutrition-related diseases.

Our data show that food composition tables do not meet accuracy requirements when the analysis involves foods that are not consumed raw, and where the presence and amount of nutrients in foodstuffs are very dependent on local conditions, mainly soil composition. This work also shows that dietary data can be reconciled with related clinical and biochemical indicators.