TD I completely agree with you that the original study did not include emergency situations or forms of cancer or rarer blood disorders. But that doesn't mean that Lee Elder's methodology isn't flawed. To have come up with a true estimate he would have had to consider a number of other factors, just a few of them:
Percentage of use of blood transfusions in the following scenarios:
Emergency Situations
Perioperative
Uses within Oncology diagnosis
Uses within haematology diagnosis
Then you would have to look at how common these diseases and forms of blood cancer are. Is it 1 in 1000 or 1 in 10,000
What is the rate of emergency care that hospital protocol would initiate a blood transfusion? For instance, in a 5 person car accident would it be necessary for all 5 people to receive a blood transfusion or 1 of them or none of them.
These are just 3 examples of questions that would need to be answered for a real estimate.
It is certainly a noble goal to come up with an estimate but what Lee Elder did was a very basic calculation without looking at all of the different variables that are necessary to make an educated estimate of the numbers.