a1 CRNH-Auvergne, BP 321, F-63000 Clermont-Ferrand, France
a2 Bio Intelligence Service, 1 rue Berthelot, F-94200 Ivry/Seine, France
Background Nutrient profiling systems aim at positioning foodstuffs relative to each other according to their contribution to a balanced diet. The accuracy and performance of methodologies are still debated. We present here a critical analysis of the structure and efficiency of the current schemes.
Methods The literature survey detected only four systems addressing the issue on an ‘across the board’ approach and with enough detail to enable analysis. The building principles of these systems were compared and their performance was estimated via their classification of a series of 125 foodstuffs on the basis of nutritional composition. These classifications were compared with one another and with an empirical classification by expert nutritionists.
Results All systems gave a similar overview, with fruits and vegetables ranked as the most favourable foods and fatty and sugary foods as the least favourable ones, but numerous discrepancies existed in every system, mainly related to their choice of nutrients and thresholds. The FSA scoring system seemed the most consistent approach, although it still generated some questionable rankings. Expert classification did not clearly validate any scheme, and cannot be considered as a true reference.
Conclusion Nutrient profiling systems are confirmed to be powerful tools to translate nutritional information related to the whole diet into the level of individual foods. However, the performance of the existing schemes remains moderate. Alternative approaches, such as considering food categories or introducing more stringent validation steps by a panel of expert nutritionists, could be ways to reach more efficient and consensual tools.
(Received July 25 2005)
(Accepted June 06 2006)