Saturday, January 31, 2009


Suddenly, an awful wet crunching and groaning and sick heavy breathing. It's...huge...festooned in the rags of a once-respectable suit, waving a bladeserver torn from a rack like a child's toy...dripping with stale blood. No! The NHS IT Zombie has escaped, and it's fortified itself by eating BT's brains. Now it's coming for us. DAATA! it groans. WAAAANT YOUR DAATA! Run!

Seriously; BT has recently had to spook the stock market by warning of a huge hit to profits from its Global Services big-IT division. But reading this FT story carefully, it seems that a lot of that or maybe even the whole thing is down to the NHS National Programme for IT, and specifically the London Region patient management contract. (The other bits are the ones that haven't gone to ratshit yet.)

The regional patient-management segments were always the most challenging bits of the NHS NPfIT; partly this was natural, because their function - a workflow, documentation, and management information system for the entirety of a major hospital's operations - was by far the most complex in the project. The NHS National Network is a big VPN; the Spine needs to authenticate users, validate input, write to the DB, synchronise, and retrieve; but the patient management system needs to deal with all the possible pathways patients take through the hospital.

Partly, however, it is unnatural and caused by the politics of the project. The regions don't actually correspond to any organisational entity in the NHS - they exist only for the IT project. They therefore have to replace existing systems that vary widely inside each region and cope with organisations in different chains of command. And each region was originally meant to be implemented by a different company; now, most of them have either given up or gone bust, and BT is doing much more work than previously planned, and this of course means that it has to deal with radically varying solutions already installed.

Worst of all, though, the regions mainly exist because the Government wanted to have the job done by the Big Consultancies - Accenture, EDS, and friends - that it was used to dealing with. Assuming that they wouldn't be interested in small contracts, the Government invented a completely new organisational level in order to sweeten the deal. They further insisted on the contracts being covered by intense secrecy, which cut off any possibility of talking to the users. And the Big Consultants proceeded to move the actual development to the US and India to save money, thus avoiding any institutional knowledge that might somehow have seeped in.

Now, it looks like BT is planning to offer a "more tailored service" to the hospitals - which sounds a lot like "doing the requirements exercise we should have done back in 2001". Of course, it's going to cost money and nobody knows how much yet, but I suppose it's progress, especially as the sacking of Fujitsu from the project means that it looks more and more like a BT job (London, the ex-Fujitsu South, the national projects, and perhaps more besides).

But it's still not too late to take radical action. Part of the original plan involved using a common data exchange standard for the whole NHS; if this exists, there's no need for much of the rest, especially not the regions and possibly not the Spine. We could define some goals and a set of data formats, then break out the cash to the individual hospitals, trusts etc to use themselves. In fact, when various US, Australian and Finnish hospital sysadmins tried that, they came up with the best healthcare IT system yet. The problem with the NHS NPfIT is quite simply that it didn't listen to the bureaucrats.

Which is why this is sense. I have no idea what such a sensible set of ideas is doing in George Osborne's in-tray, and I suspect the Tories may think it's a way of preventing IT development in the public sector. But I think a cross-government requirement for common data standards, as much open source as possible, and perhaps even building everything with a sensible API for further development would do nothing but good. And perhaps the project cap might help - after all, the way to deal with zombies is to destroy the brain.

No comments:

kostenloser Counter