Variability in the intercomparison of food carotenoid content data: a user's point of view.

The availability of reliable information on food composition is essential both for the evaluation of diet and for nutritional research to relate diet to health or disease. In this article, we compare the total and individual carotenoid contents and the retinol equivalents in fruits and vegetables reported in several food composition tables and HPLC studies. The impact of the variability in carotenoid intake was evaluated on the basis of Spanish National Consumption Statistics and on the values for a standard diet. We identify, from a user's point of view, errors concerning identification of the items and the terms used to refer to the compounds. Food composition tables overestimate (by 2 to 48%) the retinol equivalent intake, whereas they underestimate (by 30 to 50%) the total carotenoid intake according to HPLC data. We study the effect of these main sources of error, their impact on dietary assessment and on the classification of relevant contributors, and the possible consequences with respect to proper diet in terms of nutritional assessment and epidemiological studies. Given the different dietary habits among populations and the fact that certain items may be over- or underestimated in databases, the use of a single database may be misleading as to the rate of carotenoid consumption and the "true" nutrient intake in a given population, thus weakening the reliability of the study and resulting in erroneous conclusions.