You cannot really anonymised records like these. You can identify most people from only their post code, gender and date of birth (using just public data sources). So stripping names out is meaningless...
You're really underestimating the depth of anonymization in medicine. Birth dates are considered PII. Treatment dates are PII. Record numbers are PII. Locations "narrower than a state" are PII. Even ages over 85ish are PII because there tend to be very few people that old at a particular facility.
I mean, you can always anonymise data by removing all the useful data.
And that is the problem here: how do you remove enough data to make it NOT personally identifiable (or close too) AND not remove so much data that the whole thing is pretty useless.
No one has really managed that yet. People who have not tried assume it is possible. But it probably isn't except maybe in very specific cases where you only need very limited data and don't care about correlations with other factors...
There are some quite high profile examples of orgs releasing anonymised data and people linking it back to the individuals:
Interestingly the UK (I am a limey brit) actually has some really good experience with this, both from NHS medical records and public studies on Civil Servants...
