Résumé :
|
[BDSP. Notice produite par INIST-CNRS evr2dR0x. Diffusion soumise à autorisation]. Objectives : To develop a Healthy Food and Nutrient Index (HFNI) and to relate this index to the all cause mortality observed in two Belgian communities. Design : Using the 10-years follow-up mortality data from the'Belgian Interuniversity Research on Nutrition and Health' (BIRNH) study, the association of a Healthy Food and Nutrient Index (HFNI) and all cause mortality was investigated in 5,887 men and 5,306 women aged 25-74 years. The HFNI was developed on the basis of the national dietary guidelines. The higher the index the more the diet can be considered to conform to the recommendations. Results : After adjustment for classic all cause mortality risk factors, HFNI was related with mortality for men. A decrease in risk of death is noted from Quartile 1 to Quartile 4. Among the women, HFNI remains non-significant. The ability of the models (with and without HFNI) to discriminate the living and the dead was evaluated by constructing a receiver operating characteristic (ROC) curves. For all models, areas under the ROC curves were between 0.80 and 0.85 indicating that all models discriminate well alive subjects of deceased. However, comparison of the areas under the ROC curves of the models without HFNI with the ROC areas of the models with HFNI showed no improvement in our ability to discriminate between the living and the dead. Conclusion : Although it is not demonstrated in our study, we think that the overall approach offers an opportunity to improve our understanding of the role of diet in the risks of morbidity or mortality associated with the overall dietary pattern rather than some individual foods and nutrients.
|