I assume they can’t just use bomb calorimeters; they likely derive it from tables of ingredients to some extent but that doesn’t actually answer the question so much as shoves it across the plate a bit.
Bomb calorimeters are not as accurate since even indigestible things like fiber is flammable. I think the most accurate methods use enzymes to mimic digestion.
I don’t think they need to burn the food in its entirety (which, as mentioned, will also burn indigestible parts); if you know how much fat, protein and carbohydrates there are and the calories given off when they are burned, then you know how many calories are in the food.
Actual calories delivered to the body may be different yet as some foods are not completely digested above and beyond the fiber component … but that is another issue.
I would starve someone till they had no more poop. Then make 2 indentical dinners. One I would burn and the other I would feed the subject. When the subject took a poop I would burn his poop and subtract that from the dinner I burned.
Are the calorie counts provided by the manufacturer subjected to checks by gov’t regulators? And, of course, I mean more than just checking the manufacturer’s arithmetic for the components given. In other words, are the percentages (fat/carbs/protein) verified? Likewise, is the fiber content confirmed (since exaggerating its amount would permit a significant subtraction of the weight to be deducted from the ‘net’ calorie count)?
I’ve always wondered about this. What percentage of packaged food products actually get inspected and analyzed as a matter of routine? It’s one thing for the numbers to add up on paper, but who’s going to know if the relative amounts they put into the package are changed? What if, for example, 5 grams of fat secretly get exchanged for 5 grams of carbs? Simply weighing the package isn’t going to tell you what changed, but the total calories will be different.