The National Health Service is about to embark on its largest-ever IT investment. About £5bn will be spent over five years. Senior NHS officials are confident of success, but then so were civil servants when they planned the Libra project for magistrates courts, which cost nearly three times more than expected and has never...
delivered its core software. Tony Collins examines whether the NHS is learning from history.
Determined to avoid another IT disaster, civil servants in London made sure their next computer project was very different from their previous failures - they gave it a whole new name.
The previous two projects had been encumbered with the title of Magistrates' Courts Standard System. Much brain-storming produced the catchier name Libra for the third project.
Meticulous planning went into Libra, which was a scheme to give 385 magistrates courts a standard national system; but in the excitement to award the contract to Japanese-owned Fujitsu, several things were overlooked, two being size and complexity.
The previous failed projects had become mired in complexity, but they had cost the taxpayer only £10.4m. Libra was to cost £184m.
In 1998, when the Lord Chancellor's Department awarded the contract for Libra, its civil servants believed, like some of the most senior officials in the NHS today, that size did not matter as long as the risks of failure were transferred to the supplier. They were wrong.
Libra has turned into one of the government's biggest disasters. Costs have risen from £184m to £390m; and instead of having new software to manage court cases, handle accounts and help collect fines, staff in magistrates courts will, in future, have to rely on enhanced legacy software that is more than 10 years old.
Barrister and arbitration specialist Edward Leigh, who is also chairman of the House of Commons Public Accounts Committee, said, "The Libra project is one of the worst IT projects I have ever seen. It may also be the shoddiest Private Finance Initiative project ever. This is a shocking waste of money."
Leigh's words may be tinged with overstatement, but other specialists who last week read a report on Libra by the public spending watchdog the National Audit Office thought his comments restrained.
In theory taxpayers were protected from failure because Libra was signed under the private finance initiative, the main principle of which, according to the government, is that the private sector puts up the money and takes on the risks of failure. In return the taxpayer pays regular instalments which cover the supplier's costs, bank interest and profit. But with Libra, the main risks ended up with taxpayers because the Lord Chancellor's Department was unable in practice to cancel the contract.
Magistrates were, after all, in the midst of receiving new office automation systems from Fujitsu. If the contract were aborted, the department expected that the roll-out would stop; support for the national systems would be minimal; and there could be disruption to the work of magistrates courts. So the department agreed to regular renegotiations of the contract.
"They [officials in the Lord Chancellor's Department] allowed Fujitsu to run rings around them," said Leigh.
The National Audit Office report said one of the main flaws in Libra was that it sought to computerise before questioning existing business processes in use at magistrates courts.
Andrew Davies, visiting professor in information systems at Cranfield School of Management who studied the Libra project, said, "IT applications support the operation of business processes. Standard IT systems need standard business processes."
But nobody appears to have noticed that the NHS is about to make the same mistakes as the Lord Chancellor's Department. The NHS is about to spend billions of pounds on a series of national IT projects without questioning existing business processes in thousands of hospitals.
"Hospitals cannot agree on how long electric doors should stay open to let patients in," said one NHS IT specialist who did not want to be named. "In one hospital the doors knock patients over because they close too quickly.
"If hospitals and suppliers cannot agree on how long electric doors should stay open, how are suppliers and the Department of Health going to succeed with enormously complex national projects such as electronic patient records and electronic booking systems? There are thousands of hospitals, each doing thousands of business processes in different ways. You can't plant a national system on top of that and expect it to work."
The Lord Chancellor's Department said it had no mandate to force magistrates courts to change or standardise their business processes. It expected courts to adapt their business processes to the system, as it was being installed.
But the National Audit Office suggested that the department should not have signed a contract for a national system without a mandate to change business practices.
"IT system changes should be planned to support redesigned business processes. Undertaking one without the other is unlikely to deliver value for money," said the audit office.
But the NHS will soon be in a similar position to the Lord Chancellor's Department in 1998: ministers are due to agree contracts for national systems, awarded to a handful of "prime service providers" without the Department of Health's being able to force any standardisation of business practices in hospitals or doctors' surgeries.
Leigh warned that the NHS must heed the lessons from the Libra project and not try to computerise nationally without standardising business processes. The Libra project was, he said, fairly straightforward compared to the "infinite complexities" that could ensnare national NHS schemes.
If the lessons from the Libra disaster go unnoticed by the NHS the losses for taxpayers could be staggering: the government expects to spend an additional £5bn on new IT systems over the next five years. Indeed tax rises come into effect in April to help pay for extra spending on the NHS.
Officials at the Department of Health seem confident they can manage their proposed mega-sized contracts with suppliers. The Lord Chancellor's Department was also confident in 1998 in its ability to manage the Libra contract with Fujitsu.
Yet Fujitsu raised the price of the Libra contract even before the final deal was signed.
The increase came after the company's board said it was no longer able to support the charging basis on which the bid was submitted. "ICL [Fujitsu] had not included all the costs needed to meet the Libra requirement and had made some inappropriate cost and revenue assumptions," said the National Audit Office.
The department had little choice but to accept Fujitsu's new price of £184 against the initial bid of £146m. For, by December 1998, the supplier was sole bidder, and the department felt that nobody else would be interested in the contract if there were a second competition. So the contract was signed.
Only 10 months after signing, in October 1999, Fujitsu formally sought a further increase in price, this time because its forecasts of cashflow showed a deficit of £39m over the life of the deal. "ICL [Fujitsu] said that it would be unable to continue with Libra if this gap could not be closed," said the auditors' report.
Fujitsu also set deadlines. It wanted renegotiations to be completed within five months, by 21 March 2000, or it would have to declare a loss in its annual accounts. "It needed, therefore, to make a decision on whether to walk away by that date," said the report.
The department hired Ernst & Young to see whether Fujitsu could still deliver the contract. "Ernst & Young concluded that the position was worse than Fujitsu had declared: 'ICL's [which changed its name to Fujitsu in April 2002] financial model contained major flawsÉ and could not be relied upon for making business decisions'." But consultants from PACE said the project was too strategically important to allow the supplier to default.
By May 2000, 18 months into the contract, the department signed a renegotiated deal, which increased payments to Fujitsu from £184m to £319m. In return the company agreed to provide the Libra service to courts for an extra four years and to deliver a national network earlier than planned.
But by Spring 2001, the contract was in much deeper trouble. Fujitsu faced losses of up to £200m even at the higher, renegotiated price. The company said it would walk away from the contract unless the department negotiated to cover the projected £200m loss.
As before, Fujitsu gave deadlines for a renegotiated deal to be signed.
To make matters worse, Fujitsu was in breach of contract by the summer of 2001 for failing to deliver the specialist core Libra software to help courts manage their case files.
The department did not want to cancel the contract, however, because it needed the core software. In any event, its lawyers had warned that if the department cancelled the contract and sued, Fujitsu could have counter-claimed on a number of grounds, such as failing to make the requirements clear enough.
By 2002, ICL had quoted a new price of £400m for Libra, which it reduced to £384m on negotiation although this was not a final price.
This time the department said "no" and decided to revert to a scheme which had been proposed by the union, the Association of Magisterial Officers, about 10 years before. The scheme was the simplest of all the options: to turn the best of the diverse legacy systems in use in courts into a national application.
So any time now the department will sign a contract with STL, one of the legacy suppliers, to provide Libra's core software. But Fujitsu will still receive most of the money from the Libra project even though it is no longer contracted to deliver the core software.
In effect Fujitsu has gained tens of millions of pounds more money than it originally expected, without delivering the core software for which the Libra project was mainly commissioned.
More than 20 years ago Parliament highlighted failings of IT in magistrates courts, and still there is no national system for handling case files.
Even though the NHS has said it will not use PFI for its major IT projects, Libra shows what could happen should things go wrong. If libra was too big and complex, it is minuscule compared to the proposed national IT projects in the NHS. Libra was expected to process electronic files on about two and half million criminal cases a year. One of the NHS' proposed national systems, integrated care records service for the NHS, is expected to handle three million "critical" processes - each day.
Without too much difficulty a major bank could handle such a volume of straightforward credit and debit items. But the nature of health data is different.
In a typical week, NHS laboratories and associated services will provide results on millions of tests, 1.4 million people will receive help in their home from the NHS, six million people will visit their GPs, more than 800,000 people will be treated as hospital outpatients, pharmacists will dispense about 8.5 million items on prescriptions, and the list goes on.
The NHS has hit problems with centralised IT before. When ministers drove a project to computerise a single region, the Wessex Regional Health Authority, it ended in disaster and losses to the taxpayer of up to £63m. Another grand venture, for hospital information systems, which were built locally but managed centrally, lost nearly £100m in the 1990s.
If Leigh of the Public Accounts Committee thinks the Libra project is one of the worst IT disasters he has seen, he should perhaps prepare for what could happen in the NHS. If Libra costs can rise nearly 300% without the main software being delivered, it does not take a mathematician to work out what could happen to the £5bn NHS ministers plan to spend on new IT projects.
The Libra project: countdown to disaster
Early 1990s: Price Waterhouse tries and fails to deliver a national system for magistrates courts. Taxpayers lose £5m on the project but recover £1.4m from the supplier
Mid-1990s: the Lord Chancellor's Department hires suppliers including Admiral and FI to help build a national system but the contracts are terminated in 1996. The department wastes £6.8m
Summer 1998: third attempt at national system. Fujitsu submits a bid for £146m, is the only contender for Libra, and is appointed preferred supplier
December 1998: Fujitsu increases its bid price to £184m. The department agrees and signs a PFI deal based on a system specification called Mass, which was used in the previous failed contracts
1999: Fujitsu seeks a contract renegotiation as its cashflow forecasts show a deficit
2000: A revised contract is signed, giving Fujitsu £319m
2001: Fujitsu's forecast losses are so high it says it cannot continue on the contract unless it receives an extra £200m. The supplier is in breach of contract for failing to deliver the core software
2002: Fujitsu proposes an increase in the contract's cost to £400m, a 273% rise on the original bid price of £146m. The department says no but signs a further revised contract giving Fujitsu £232m and relieving the supplier of responsibility for delivering the core software. The new money will cover new PCs, automation software and a national network
2003: The department is due to sign a contract for the core software to be delivered by a legacy system supplier STL. A third company, not yet appointed, will provide systems integration. The current projected cost of Libra is now £390m.