It's also just a lot of legacy stuff that no one wants to touch because it currently works and brings in money. Good design and practices compared to today just weren't around 10-15-20 years ago. Building a large and robust EHR and/or practice management solution from scratch using good techniques and design (and good security practices like, stupidly enough, parameterized queries instead of string concat) would be great if it wasn't astronomically expensive and risky from a business standpoint. Then by the time you were ready to hit the market you'd be five years behind or something.

Plus everyone will ask "Can you convert some other EHR's notes/data into your system's notes and vice versa?" and now you get to inherit all of that system's bad decisions in that area.