McDonald's plan to get people to understand them is to put the words next to the icons to get people to understand the icons, which also wouldn't help if you can't read.
Sure, only a relatively small percentage of the population of Germany/Poland/etc may not be able to read. But how many of those visit McDonald's? Do they visit it more or less often than other categories of people? How important is it for them to be able to understand these icons? Is the cost of developing icons/signs they also understand more or less than the amount of trouble them not understanding these icons would bring? Are we talking about actual literacy or the more nebulous category of "functional literacy"?
There are cases where "devil's advocate"-style questions are worthwhile, and I'd argue that they're worth at least considering in this case.
For example, for proteins: the word "protein" is understandable in English, French, Spanish, Italian, Portuguese, Greek and Swedish. The Russian "belok" should be understandable to Polish and Russian speakers, and the Dutch "eiwitten" should be understandable to Dutch and German speakers. Thus, writing "protein/belok/eiwitten" would be understood by almost everyone. "kcal" is even simpler and should be understood by almost everyone everywhere - maybe make it "kcal/calories" for more readability. "salt/sol/αλάτι" should cover everybody as well. Perhaps throw in the Cyrillic as well if McDonalds expects a large amount of uneducated Russians in their restaurants.
You could even put the native language in front to prevent any insult to cultures. Am I crazy, or might this actually work?