NHS IT: what went wrong?
Accidenture got involved!
'Nuff said.
This week Fujitsu pulled out of the £12.7bn National Programme for IT - the government's enormous project to introduce national patient records for NHS patients leaving BT, CSC and iSoft still involved. How bad must things be for a company to walk away from just under £1bn of government money? And why are we writing yet …
G. Brown "Its working as indended, now a Dr can know about someones in-growing toenail anywhere in England, in the past they would have to phone the patients doctor which took to long! and this whole system only cost X Billion pounds....the tax payers should be grateful for this Bargain"....
"Data exchange" isn't hard. In the worst case scenario, patient records can be printed out and sent by fax or emailed. There's more manual overhead involved here, and always the chance that the updated records won't be sent back, but this is what the health service are going RIGHT NOW and although it's not perfect, it's hardly a crippling problem.
As an ex-civil servant, I can give you a few reasons why government IT projects will nearly always fail in some way. We used to pad our costs estimates by a minimum of 20%, as we knew that whatever budget we requested it would be cut by around that amount as a matter of course. Then there was the usual scope creep, with no additional money for additional features. Failure to consult with all stakeholders, not just senior management, was common. Not because the peons doing the work didn't think of it or couldn't be bothered but because we weren't given access to the lower grade people as their views weren't deemed important. Lowest quotes always got the job if they could just about manage the mandatory requirements, rather than evaluating true "value for money" and paying 10% more for 50% more functionality or service. Costs would spiral partly because of missed requirements, but also because of headcount rules - contractors on a project rather than BAU could come out of the capital budget rather than the current one, whereas all in house staff were from the current budget.
You'd also get issues relating to the fact most politicians just don't live in the real world; you'd be told to put together a major mainframe system in 6 months from scratch because of legislative changes, or to produce an important system with high functionality for peanuts using a small PC network.
Add to this that most of us were paid significantly under the market rate for the skills, and that we were treated like naughty scoolchildren rather than responsible adults, and people wonder why the staff don't seem motivated. Most of us did our best, but demoralised staff working to absolutely impossible deadlines and budgets rarely results in a fully sucessful project!
Coat icon, because I got mine & left the civil service years ago when i couldn't take it any longer.
Its interesting that The Reg plays up that Fujitsu pulled out, while other reports suggest that CfH (Connecting for Health, the NHS central IT body) "terminated" their contract.
Not many in the NHS are weeping over Fujitsu's departure. They started off with an attitude, something like "This is what you are getting, take it or leave it" and went downhill from there. The product they were pushing, which had been agreed between them and the DoH, without any involvement from the poor s*ds who would have to use it, was simply not fit for purpose. When Trusts pointed this out, they were threatened with financial penalties if they didn't shut up and get on with it.
I don't know who to blame for this shoddy mess, but I do know, from personal experience, that Fujitsu were difficult, unhelpful, arrogant, ill-informed about the NHS, and will not be missed.
The only downside is that now we might get lumbered with BT. Out of the frying pan...
From teh Dome to the Library, to this and that and then to NHS, almost if not all are a total fcukup, I am now waiting to see the final results of the Olympic bill and the yet to come ID crad bill, no doubt ElReg will have more to write about soon.
When will they learn that the gov cant organise shit ??
I don't know much about this particular project, but my experience of working on NHS IT projects is that different Trusts, Hospitals, departments, etc. all have different ways of working, but, since managemnet wants to extract comparitive metrics from the system you can't have different systems for each one. So, in trying to accomodate them all you end up with an overly complex system that borders on the unusable and unmaintainable.
The alternative is to have a reasonably straightforward system, and demand that everyone changes accordingly, but this would require unrealistic changes in operational practices, most of which exist for good reason.
Perhaps, we should have systems that do a simple task well, instead of some megasystem that tries to do everything, and ends up failing.
"In the private sector it has to a mega mess before anyone hears about it."
Not strictly accurate. In some private sector areas (tech journalism for online magzines; to take one completely random example) there is always the chance that a pedant with nothing better to do with their time will spot a tiny, insignificant mess and publicise it; such as noticing the word 'be' having been accidentally left out of a sentence then posting a comment about it where literally dozens of people might read about it..
In a word iSHAFT, as they were affectionately known to staff when I worked there.
To the AC, in fairness to Accenture (also worked there) they may have had failings, but were royally shafted by the aforementioned software provider (who failed in the most basic of requirements, that is, to provide software). That said, they did over pay their contractors -- which I was most upset about :o)
there goes my bonus :(
in all seriousness though, Fujitsu are taking enough flak over DII/F because of other parties that are involved being less then ideal partners and i think it was a wise business move to jump rather then sink with the ship so as to speak.
black helicopter because thats the closest you have to a representation of a company thats helping wage war on poppy farmers.
Customer wants X
Government wants Y
Supplier wants $
Supplier offers Y Sells Z and Manages Expectation of X
Customer Gets 0
Government Gets 0
Patients Gets a leg amputated instead of a hernia op.
Then We can't find the patient needing the leg amputated and is left
waiting in a lift, rotting nicely.
Supplier Changes Name.......Again !
Ministers Get Seats on the Board.
Share prices Increase.......
Whats New !!!!
There are quite a few obvious reasons why the National Programme didn't move in the direction of your "expert" contributors:
1. Simon Williams, director of independent consultancy DMW Group, said: "They should have started small - run a project in one or two small regions with one hospital and two or three GP surgeries in each. Run a prototype project, prove the functionality and show the staff the benefits of it - it is a simple and well trodden route for big projects."
Mr Williams clearly has limited experience of working in the NHS if he seriously believes that a pilot in one localised area would actually prove the benefits for a much wider rollout. He suggests maybe taking one hospital - but what type of hospital would be representative?
Should we pick a district general hospital, regional specialist centre or a national centre? What about the services that these hospitals don't provide but others do? How would we factor in that?
Could we really demonstrate that what works in one hospital would work in another down the road where the internal processes and legacy systems are completely different? And how can we be sure that what works in a district general hospital would work in a much larger hospital? The same argument applies to GPs with the mix of single handers, small partnered practice and major polyclinics like the newly opened one in Hounslow.
"Williams added that having proved a small scale project works it can, if necessary, be handed over to a bigger supplier to roll-out nationally, once the functionality has been proved."
...and misses the key point that the project isn't just about banging out new systems, it's about new systems that support many and varied ways of working that even most clinicians in the NHS can't agree on.
2. "Martyn Hart, chairman of the National Outsourcing Association said: "If I live in south London how many times will I need my health records to be sent somewhere else? How many times does that really happen?
Well if Mr Hart gets referred to one of the 4/5 providers on Choose & Book who might happen to be outside his locality he might very well like the Doctors there to see his notes. And if Mr Hart is unfortunate to get referred for specialist treatment e.g. for Cancer at the Marsden (so not in South London,) the same logic applies. And what if Mr Hart were to be in a serious accident or taken ill outside of London? Wouldn't he want the Doctors to know that he's potentially allergic to commonly used drugs that might just kill him before they start injecting them?
" you need some agreement on formats and standards not necessarily a nationwide system.""
This is precisely the model CFH have adopted - defined standards that multiple system providers have to support - thereby delivering a standards based approach whilst retaining contestibility amongst suppliers.
As ever, these "experts" demonstrate little if any appreciation for the scale of the challenge or the problems that CFH have to grapple with on a daily basis. It's very easy to pick holes in something when you don't have to come up with a plausible alternative that would actually work in the real world.
Declaration of interest: I work with CFH for a supplier and have seen just how complicated / ambitious the National Programme is - and the considerable amount of effort that goes into simplifying solutions whilst retaining the use related benefits.
Hmm, I've long wondered to myself how I would work a project like this. Shoot me down if this seems silly but ...
1) I would require all the design documents and code to be open - placed in the public domain. The contractors are being paid for the costs + profit from the public purse, so making all their work open for public scrutiny should be the norm.
2) I would set much less ambitious "in service" deliverables for earlier in the project. Point solutions where the data can be federated (and later maybe centralised) stand a far better chance than putting in place overarching mega-infrastructure and then insisting the apps conform.
I've never been able to see why public sector IT should be based on companies creating proprietary and closed solutions which are resold again and again. Why not pay them to create free (libre) solutions and then pay either them or other companies to implement them well?
I've been working for a supplier of ... a supplier of a ... into CfH for a while now and I have to say I'm pretty stumped as to how the Government could possibly have succeeded here.
In the end, the functionality they're looking to build isn't that enormous. Ok, it's pretty complex, but the real difficulty with the NHS is the size of the damned thing. It's the third largest employer in the world (Chinese Army=1, Indian Railway=2) with over 1.2m employees when last I checked. That's big. Creating ANYTHING for an organisation that size is a massive undertaking.
The article takes the glib line that if you start with a small pilot and then roll things out further it's all much easier. Well, guess what: of course they start with a small pilot and roll things out! What kind of idiots do you think these people are?
The problem is that a system built to work well in one trust with one set of supplier systems to integrate with, one set of hardware (Windows 98 anyone?) and a small number of pilot users simply doesn't translate into national roll-out. Data exchange isn't easy - it's damned hard when you've that many different systems, and that many different suppliers each looking for a competitive advantage by forcing their features into the data formats.
The actual problem that all the LSPs are facing is that the suppliers (Accenture, CSC, Fujitsu, BT) were made responsible for delivering the technology and had contracts that said they would be paid based on uptake. However, it was specifically out of their remits to manage organisational change in order to ensure that uptake. This was a setup that could never succeed for them. Accenture pulled out when they realised they could never succeed, Fujtisu seem to be the next to realise the same (contract terminated means "unless you change the contract so we don't get screwed, we're quitting"), BT are only really interested in the networking piece and are half-heartedly doing the healthcare bit as an afterthought. CSC seems to be the only one that's doing ok, and that's largely because they've more or less ignored the GPs and stuck to the hospitals, where at least you can handle some of the change management by sticking ads up in the cafeteria...
The real difficulty is that what they're trying to achieve is actually very worthwhile. Microsoft HealthVault, Google Health, all this stuff - it's the signpost for where the world is going, and electronic patient records are a key part of it all. In the end, I guess it comes down to whether we want the NHS/Government to control our health data or should they just give up and sell us all to The Beast(s)..
</rant>
At the end of it all, the consultants and Lawyers are still getting paid and getting rich off all these failures. To them, this is all just business. It doesn't matter if the customer is delayed by their direct clients - that's not the consultant's fault. They only have to get paid *again* to vet and approve a new client for the customer to sign up with.
It a really shoddy state of affairs and the whole system needs to be re-done. This is about the civil-service, rather than government - and that's as hard to change as changing Paris into a Lady.
...knew there was a Paris angle somewhere!
It is possible to infer how much pain Fujitsu absorbed before walking away by noting how tight-lipped they have been since the announcement. They wouldn't walk away from a revenue stream that broad and deep without severe provocation.
The official line in notably terse. To paraphrase: "The situation has developed not necessarily to Fujitsu's advantage".
"1) I would require all the design documents and code to be open - placed in the public domain. The contractors are being paid for the costs + profit from the public purse, so making all their work open for public scrutiny should be the norm."
How would that improve the code? Because everyone can see any flaws? Great - watch the gillion emails from Reg readers who could do so much better roll in every day and swamp the coding companies. Or do you think just anyone should be allowed to chip in? Do you think that'd work? Really?
I don't really know anything about how your second point might work in practice, but it sounds reasonable enough.
"I've never been able to see why public sector IT should be based on companies creating proprietary and closed solutions which are resold again and again. Why not pay them to create free (libre) solutions and then pay either them or other companies to implement them well?"
A software company makes money from its software. Why let other people get that money? Lofty open source ideals don't pay the bills. This open-source-solves-everything mindset is pretty flawed, I think.
I think you're slightly avoiding the real issue.
The problem is the fact that there is supposed to be a national rollout. This implies that the final software is one homogenous lump that runs in every location. Of course this would cause problems. The final aim is to have access to patients data, not to have the same software on every machine.
The project should concentrate on getting one district working with data being collected and stored in a uniform format. The next district can have any other proprietary form it likes as long as the data collected is stored in records with the same names. There is bound to be substantial overlap in the records anyway, as patients usually keep the same name, DoB etc,etc.
The main DB stores all the records from all the districts by containing all the necessary fields, ie. it collates the data, so the query engine accesses the main dbase and so has access to all the fields. No need to force different procedures on any particular district, and no need for a national rollout, other than to build the query engine for data access, which can be updated as and when needed to add new queries.
Surely this system is going to be networked, so using proprietary software to create conflicting storage formats should have been disallowed in the original specs. When I upload a file from win98 to my linux server it matters not that the origin was win98. Just force the format of any uploaded data to be one that your end user query engine can understand. You do this through the original spec.
I would have people going to every hospital in the first instance, and documenting every field in every patient dbase. Then you take that data, and collate it to optimise what you actually need to cover all the bases. Then you go back to each hospital and make sure that their particular system can talk to the main dbase using the optimised field structure. They can keep their old software, just add a new layer that queries their system. The new layer talks to the main dbase.
Of course this doesn't allow any one software house to dominate the national dbase, and therefore is worth less money, but what was the original intention again ? Oh yeah, better access to patients data. Well let's start with that shall we ?
Degenerate Scumbag is right! A database this expensive that covers every person in Britain should be made to cover its own costs by charging something to the beneficiary, that is, every person in Britain, and that's why we have taxes. So that's OK then. Next we should have the greatest possible efficiency of use, so throw it open to scope creep into the National identity Register. And that cool new idea of recording every phone call and email. And the road-charging. And the ANPR cameras hanging off every motorway bridge. And HMRC's list of all of us too. Don't keep repeating data, just have one master version of each. Private businesses have one staff register and tie salary and disciplinary info to it, why can't the whole UK? The Ministry of Love can look after it for us.
I'd be curious to see exact details of what is being done. I mean, how hard can this be? You've got patient name, address, age, place of birth, nationality, sex etc, and then you've got details of their current and past medical conditions.
How hard can it be to get this data into a single database accessible with a simple resource light, multi-platform interface?
My suspicion (going by observations of how NuLabour works) is that Blair/Brown want to create a system that effectively micromanages the whole of the NHS. A single recording structure that tracks everything from drugs prescription costs to heart transplant failure rates would be very useful to a control-freakish central government.
Maybe this system is more about monitoring how the NHS is behaving (down to the level of the individual doctor) and ensuring that performance and budget targets are met than making sure patient info is easily accessible to healthcare workers. NuLabour clearly believes that weighing the pig makes it fatter, and the NHS is a very big pig.
And I bet that the Fujitsu contract cancellation (and the drop-out of two potential ID card scheme providers) has come about because Brown is trying to cut costs to cover the whole 10p tax band cock-up and other budgeting catastrophies. Brown reminds me of a boss I once had who seemed to believe that all a big spending consumer (like the Treasury) needed to do to gain extra cost reductions was to dig his heels in and tell the company that if they didn’t do what he wanted then he’d go elsewhere. The fact that what he wanted was often impossible at the price he was prepared to pay was irrelevant. After all, market forces are supposed to ensure that the consumer always get what he wants!
For those that think that they didn't consult GPs etc, the Scottish Government used to sponsor (may still do, it's a while since I worked in health) a system called GPASS for running GP surgeries. Given free to all surgeries in Scotland, written by GPs for GPs, did everything a GP could need done, and yet it still had only 80% of the market because some GPs insisted on paying for systems from other suppliers that did more or less (often less) the same thing. It remains the problem that unless you mandate a solution, in an organisation the size of the NHS someone will find a good reason for using something else that defeats the purpose of standardisation and removes any economies of scale.
the govt. wants this system so it can implement billing, so NuLabs mates from US Health can get a slice of the action.
It's being sold to the voters on the basis that when you get cut out of the wreckage and taken into Luton & Dunstable General the A&E people will be able to find out you're being treated for an ingrowing toenail at your home location.
2 problems, billing only required if you want the US to fuck up the NHS and the A&E people need to get you breathing, not start fucking about with computers.
Just use a bulletin board. Separate forums for counties and each person represented with a topic of their NI number. Just post the medical records and update as necessary. No need to worry about security- it's going to get hacked anyway, no matter how much is spent.
I'll just take £1 million for the consultancy, thanks...
"the govt. wants this system so it can implement billing, so NuLabs mates from US Health can get a slice of the action."
nice idea but you clearly have no idea what you're talking about. Trusts already "bill" for activity via Payment by Results. In addition, HES data can be used to determine overall costs to commissioners.
Maybe if people stopped coming up with ridiculous "Nulab" conspiracy theories we might be able to have an informed debate.
Sorry, but paper records and overnight delivery on request just won't cut it any more.
The formats for data cannot be settled for all time because (1) new science will lead to new dimensions of data; and (2) the semantics of what will be diagnostically significant in the future cannot be known in advance.
Even without emergency incidents, we need secure federation of health data. Imagine:
My orthodontist needs a high-fidelity copy of X-Rays taken by my dentist.
My dentist discovers she needs my blood pathology results.
My rheumatologist finds he needs my parents' genetic testing results.
My motor insurance becomes invalid if my psychiatric clinics are irregular.
The Centre for Disease Control needs to identify every carrier of a genetic liver disease who is likely to have been treated with a particular topical anaesthetic.
Imagine the cost of re-creating this data, if it is not shared. The personal consequences of inappropriate release of the data could be terrible. The need for these specific exchanges were not known when the data was created. No-one seems to remember their whole medical history. Providers go out of business and old records may not be recoverable. The population is ageing and the scale of healthcare services needed can be expected to rise. These issues have been well canvassed in the Health Informatics community. The competence of the NHS may be criticised, but no other organisation (in the world) has within it sufficient knowledge of the relationships between British health service providers.
Perhaps the only way the current SNAFU could have been avoided was to limit the first stage to deliverables which would produce measurable benefits within one year.
Several friends working on this project in the Portsmouth / Southampton areas and they say that the problem is not necessarily with the key contractors but with the quality of project/programme management executed by the NHS.
The primary problems seem to be extremely bad day to day (i.e. basic) programme management, zero "buy in" from the key NHS project stakeholders in support of the solution, and an existing NHS record system that belongs in the dark ages and is so fucked that it would take two years to get it into a shape that could even begin to be migrated.
Of course the other problem is bad NHS personnel management as most of those who will be using the system to assist patients are 40-65 year old women who just love to resist any sort of change, and fawn on every work Mr Senior medical consultant makes; and generally it is Mr Senior medical consultant who calls the shots.
Why did it fail?
Well,I haven't worked on CfH but I have people working for me who have and I've spent my whole career in goverment consulting, and in my opinion is that it was just too big and too hard.
Do the people posting really think that everyone working on CfH is an idiot and if they got involved it would be so much better? Are you all that arrogant that you think you're just much smarter than them? In that case apply for a job on the next big government IT project (ID cards?) and start eating your words!
What's the point of this £12B cock-up? I've read lots of articles and lots of commentary, but no-one seems to have stated the obvious. The point, and the only point, is to screw GPs. When the record is centralised, the GP loses control. In principle, anyone can then provide primary care - a telephone operator in Bangalore, a nurse on NHS direct, a Ukrainian immigrant working in a government polyclinic. There are various other theories on what the database is actually for, but none of them stand up to a couple of minutes scrutiny.
To the pointless Orwellian functionary in Whitehall, this is a Good Thing. The stated agenda is Patient Choice, and they're getting more of it. With more Patient Choice, the gumment can force down the price of primary care. It also has the desirable effect of reducing the wages of a group of unaccountable, reactionary middle class intellectuals who won't toe the party line, and who have the temerity to think that 10 years of medical training somehow qualifies them to earn more than a politician whose only work experience is as leader of a labour council in Scunthorpe, and whose only financial acumen is the ability to triple their salary by lying on their expenses claims.
The extraordinary thing is, that primary care currently costs pretty much zero anyway. If memory serves correctly, the gumment currently pays about £220 per person per year for their GP services. That's for *everything*, including the pay of the receptionists, secretaries, cleaners, GPs, locums, and nurses. How can you possibly reduce that, short of outsourcing to Bangalore? No wonder the system is broken. Only a complete moron, or a Labour politician, could possibly think that the answer to this underfunding is to spend £12B on a computer system. Duh.
One more thing. How many of you have actually seen a real, unsanitised GP record? They're incomprehensible, and may well be useless after someone has transcribed them and put them on a database. They're notes and letters. They probably make sense to the GP who wrote them, and might make sense if they're read by a partner of the person who wrote them (another gumment 'innovation' - you don't have a personal doctor anymore; not enough 'Choice'). My wife occasionally leaves her records on the kitchen table, so I've seen a few; mainly of elderly ladies in nursing homes, who have no intention of moving to another practice, or being knocked down outside a hospital in Scunthorpe. Oh, and she works 52 hours/week, for 20K more than an entry-level labour MP, and 20K less than me, when I can be bothered to work.
No icon, since El Reg doesn't have one of a fat arse. Or a pig walking on its back legs. Got to do my VAT now, which I've been putting off all morning... :(
"My suspicion (going by observations of how NuLabour works) is that Blair/Brown want to create a system that effectively micromanages the whole of the NHS. A single recording structure that tracks everything from drugs prescription costs to heart transplant failure rates would be very useful to a control-freakish central government."
Indeed, which is why defining standards and leaving the records themselves in the hands of clinicians or (shock horror, something the DoH has fought against since Lloyd George) patients, just won't do.
Look up the Secondary Uses Service
http://www.connectingforhealth.nhs.uk/systemsandservices/sus
- a transparent misnomer, since it immediately comes clear once you read it that 'Secondary Use' is the primary bureaucratic motivator for the whole damn thing.
Yes, it is really really hard. OFCOURSE! As many of the commentators already have pointed out. People are not necessarily stupid - but arrogant! You have to ask yourself the question: WHY IS IT HARD? The answer is not only "because of complexity". There is a very important factor hiding behind concepts as "stakeholder buy-in". OK, so people involved in the development might not actually be stupid - but in many ways I do not feel sorry for those professionals who are engaging in efforts to develop systems for "users" (other professionals) who - do not want their system! Yes it is hard, really really hard - to work in an environment where you insist to treat users as ignorant and incompetent - because it is obvious that you yourself know what is best for them in their professional role - which you obviously do not have to do yourself.
I sincerely hope that these kind of abusive project continue to fail, and also continue to be just that - "really really hard". Hopefully one day people will just give up on their patronising efforts to implement systems over the heads of those professionals who are supposed to use future "wonderful" IT systems.
Where you get to decide who controls, corrects, and has access to YOUR information? If, as R says: My motor insurance becomes invalid if my psychiatric clinics are irregular...what if your insurance becomes invalid because whoever keyed in the information got your records mixed up with a govtard at #10 and you (like those on the pinko TSA list) can't ever get it fixed?
@ Odius Grunt
"I guess it comes down to whether we want the NHS/Government to control our health data or should they just give up and sell us all to The Beast(s).."
Scuse me just one second but why should the Govt control *my* health data? I've been adivsed I cannot opt out of it - the Govt ignoring it's own laws I see. With that sort of breathtaking arrogance it's hardly surprising that people are going to walk away from such a project.