Sunday, September 16, 2007

HOWTO Build a Healthcare Computing System

Remember this post regarding the trainwreck that is the NHS National Programme for IT? In it, we discussed the institutional factors that made it a trainwreck, and specifically the way demands for commercial secrecy made it impossible to involve the users in designing the system.

Well, now there's a discussion at Kevin Drum's about how the US Veterans' Administration got a really good patient information system. It turns out they asked the users, then rolled their own. Not just that, but they kept asking the users.

There's loads of information here; so if you're a techie you can add DHCP to your mental list of confusable multiletter acronyms. As well as the Dynamic Host Configuration Protocol, it's also the Decentralised Hospital Computer Program. More importantly, you may recall I queried the need for the NPfIT's structure of local, regional and national systems on the basis that, as it also required common standards for information exchange, that meant the local systems should be able to intercommunicate without needing the extra layers. Here is a whole working example of such standards.

Further, there's a standard RPC interface to lash your applications to the database, with details of the SDK here, and guess what? It's all open source and completely free. And guess what? A bunch of Finnish hospital sysadmins built a graphical frontend for it, which is soon to be ported to Java. Or you can use Visual Basic and SQL. Python is surely not far off:-)

Seriously, I've just read through this stuff and it's just desperately cool; and nobody, literally nobody, in the UK has even mentioned it with regard to the NHS Monster Database. Despite successive Health Secretaries' much-hyped efforts to find out "what the Americans are doing", which has involved various more or less horrible forms of backdoor privatisation, they don't seem to be aware that a bunch of civil servants built a cracking healthcare data system - and it's FREE.


Sir S said...

we did this in my health care centre in Australia. It included a methadone dispensing database, and basic records on all of the clients' treatment history (though none of the written details of their medical records). It was completely compatible with the broader data needs of the area health service for whom we worked. It cost one part-time database developer for 18 months, and everyone at the centre loved it because we consulted them. My job was the planning, concept design and consultation. I am now a firm believer in localised systems for specific centres (or types of centres) based on a universal data standard. And I don't think that it necessarily needs to be done by commercial consultants.

Interestingly while we were developing our system, the area health service in which we worked was quietly going through the end stages of a 5 year trainwreck in their attempts to set up a single data system for all of their centres, using external consultants...

Anonymous said...

The fact that the whole thing is written in some legacy 60s specialist language which is practically only used in the US healthcare industry these days if it's used at all would give any manager thinking of using this system the heebie-jeebies.

Otherwise it sounds great...

kostenloser Counter