battleship, you said "I still struggle with the leap to assuming that it is the formula feeding that has led to these deaths. There are so many complicating factors - espeiclaly with pre-term babies."
Exactly - this is why you cannot point to any individual pre-term infant death and say 'formula feeding caused this'.
So what you do with the numbers is this: you get a large number of pre-term infant death statistics - lets say, oh, 10,000. In a big number like that you'll have a range of risk factors which we already know contribute to infant deaths - gestational age of the baby, degree of growth retardation, use of drugs by the mother and so on. And you isolate only one factor - whether these babies had breastmilk. Lets say you find 6,000 of them had breastmilk, and 4,000 of them did not. You find that 60 of the breastfed babies died, and 80 of the ff babies died. If the risk was the same, you'd expect 40 of the formula fed babies to die - but instead, you have 80, and that means 40 excess deaths.
(Some studies may control for whether very prem babies are more or less likely to receive breastmilk, whether growth retarded babies are more or less likely to receive breastmilk as well - to make sure it really is the feeding method that makes the difference.)
In these 40 excess deaths, no one would be able to say 'if this particular baby was breastfed, he would have lived* - but we can say, if those 4000 babies who were formula fed were breastfed, only 40 of them would have died.