> pretty easy to use statistics to determine a common food or infection source.

It's getting harder. Privacy protections mean this type of analysis across hundreds of millions of records isn't really possible anymore. Without hundreds of millions of records, identifying "people who drank milk from Mr Smiths farm in 1969 were 200% more likely to get parkinsons 45 years later" couldn't be done.

Yet if Mr Smiths farm had cows with some as-yet unidentified prion disease, this is the only way we'd find it.