Large government IT projects end up being done by companies expert in extracting as much cash from the government as possible. The end result is the product doesn't work properly and is massively over budget. Then the government is forced to hand over even more money to redesign bits to work better.

The government has been bitten by this many times, so now shys away from big IT systems. Instead every department has to build its own, usually with lots of Excel sheets and macros involved.

The move at the moment is for central orgs to specify standards and protocols so it doesn't matter who implements them and the delegated authorities can buy from whoever. That goes some way to preventing both one-big-vendor setups and unstandardised fragmentation, and it was blindingly obvious to anyone who's ever implemented more than a bash script that it was the correct approach even 20 years ago. Only it's taken a generation of civil servants ageing out at the top end to make it happen.

This is largely what Connecting for Health attempted, and the only good things that came out of it were intangible to most - N3 and the UK SNOMED extensions.

Unfortunately this was largely down to incompetent consultancies, and a focus on providing what management wanted rather than what clinicians wanted, and it has absolutely poisoned the well with regards to information technology in the NHS.