Heats of combustion of the main carbohydrates contained in plant-source foods.

In a previous review, the experiments of American chemist W.O. Atwater were critically examined, with the findings demonstrating certain weaknesses that could compromise the validity of the values currently used for metabolizable energy. An examination of published works on the heat of combustion of carbohydrates reveals 2 types of weaknesses: the inaccuracy and imprecision of the calorimetric data used, and the averaging procedure employed to estimate such representative values. The present review focuses on the first type of weakness, namely the inaccuracy and imprecision of the calorimetric data used in previous studies. An exhaustive bibliographic search yielded almost 100 heat of combustion values for some of the 6 main carbohydrates contained in plant-source foods (glucose, fructose, sucrose, maltose, starch, and cellulose). These heats of combustion were subjected to rigorous statistical analysis to propose the following for each carbohydrate: (1) an interval (termed a bibliographic interval) that very likely includes the actual heat of combustion value and (2) a "representative value" (calculated to produce the minimum level of inaccuracy). In addition, an estimation of the maximum level of inaccuracy that could be expected when using such a representative value is reported.