the focus on records is unhelpful, automation is what people should be thinking of.
Records follow naturally.
The Unix approach of many small programs each doing one thing and doing it well, but talking easily to each other, is a good one.
This is quite opposite the idea that in an ideal world all hospitals (and why just hospitals?) would adopt the same “platform”.
People get it wrong even by saying apparently sensible things such as “First decide what doctors do (not that easy) and then automate the bits that are helped that way”
The error there – which was in full flight at the beginning of NPfIT – is that a complete description is unfeasible. THe correct approach is to identify one thing done by some doctors, and provide automation for the tedious computable bits of that. Then move on to another tiny bit of the problems.
NPfIT has taken a decade out of progress. Microsoft Office adoption has taken another out.