OBJECTIVE: To evaluate the agreement between macronutrients and food energy values analyzed in laboratory and data of food composition tables and software available in Brazil. METHODS: Eleven types of food were analyzed totaling 701 samples. Student's t-test was performed consisting of the comparison between the mean values obtained in laboratory and the unique constant value of each table or software at 5% confidence level. RESULTS: It was noted that depending on the type of food, nutrient and table or software selected for comparison, there were significant statistically differences between laboratory and tables and software data. Concerning the studied types of food it was observed a trend of overestimating protein and carbohydrate content in two tables while carbohydrate content was overestimated in a third table. One of the software programs showed overestimated lipid content and, consequently, the total energetic values while a second software program underestimated all nutrients when compared to laboratory results. CONCLUSIONS: The results showed it is essential to create a Brazilian food composition table from laboratory data to assure more accurate data.
OBJECTIVE: To evaluate the agreement between macronutrients and food energy values analyzed in laboratory and data of food composition tables and software available in Brazil. METHODS: Eleven types of food were analyzed totaling 701 samples. Student's t-test was performed consisting of the comparison between the mean values obtained in laboratory and the unique constant value of each table or software at 5% confidence level. RESULTS: It was noted that depending on the type of food, nutrient and table or software selected for comparison, there were significant statistically differences between laboratory and tables and software data. Concerning the studied types of food it was observed a trend of overestimating protein and carbohydrate content in two tables while carbohydrate content was overestimated in a third table. One of the software programs showed overestimated lipid content and, consequently, the total energetic values while a second software program underestimated all nutrients when compared to laboratory results. CONCLUSIONS: The results showed it is essential to create a Brazilian food composition table from laboratory data to assure more accurate data.