Below are the results of an investigation by the National Audit Office into the troubled go-live of a Care Records Service at Nuffield Orthopaedic Centre. The Care Records Service is the main part of the NHS National Programme for IT [NPfIT]. The Nuffield Orthopaedic Centre in Oxford was, in December 2005, the first NHS trust in Southern England to go live with an NPfIT Care Records Service based on a system from US-based healthcare specialist Cerner.
Letter to Richard Bacon Esq MP
House of Commons 22 June 2006
From Sir John Bourn, Comptroller and Auditor General [head of the National Audit Office].
Thank you for your letter of 13 March 2006, in which you asked me to examine the introduction of the new care records system at the Nuffield Orthopaedic Centre (NOC) in Oxford.
A team of my staff visited the Trust to look into the introduction of this system. The team have interviewed the NOC’s Chief Executive, its Director of Nursing and Operations, the Chief Executive of the Thames Valley Strategic Health Authority (who is also the Senior Responsible Owner for the Southern Cluster of the National Programme for IT), and the Chief Operating Officer of NHS Connecting for Health. They have also reviewed documents provided by the NOC and consulted Fujitsu.
I will set out the background to the introduction of the new system at the NOC, the problems which arose when it was introduced, the report of the Serious Untoward Incident (SUI), and the action being taken to learn lessons from what happened during the introduction of the new system.
The background to the introduction of the new system at the Nuffield Orthopaedic Centre (NOC).
The Nuffield Orthopaedic Centre is an acute specialist teaching hospital in Oxford with a national and international reputation. It offers services in musculoskeletal medicine and surgery (for example orthopaedics, rheumatology, metabolic bone disease, chronic pain management, pathology) and in enablement (continuing disability management, posture, independence, wheelchairs, prosthetics and orthotics).
The Trust has had an urgent need to replace its bespoke Patient Administration System (known as NOCPAS). This had been built on site in the 1980s and was not used at any other NHS sites. The system was antiquated and reaching the end of its useful life, with only one member of staff trained and experienced in its structure and development available to support the System. In addition, its database was due to reach the limits of its capacity in early 2006 and this capacity could not be further expanded (to continue to use the System the Trust had already been forced to delete data to free up space). Critical data tables would reach capacity in February 2006, and the NOC, therefore, faced this definite “drop dead” date. In addition, development had been frozen for the last two years to protect the NOCPAS at the end of its useful life. This had forced users in some areas to develop their own standalone systems to support their needs.
Given this pressing operational need for a system to replace its NOCPAS, NOC and the Southern Cluster of the National Programme for IT agreed that the NOC should be the first trust to take the National Programme systems supplied by Fujitsu. This was on the basis not only of the “drop dead” date but also because NOC is a relatively small trust, mainly serving elective patients, without, for example, Accident and Emergency or maternity departments, and so would represent a suitable pioneer trust for the new system.
To this end, Fujitsu has had a project team at the NOC since early 2005. The NOC chose to implement the system in late December 2005 because, as it is a primarily an elective site which deals with booked in appointments and does not provide, for instance, accident and emergency services, it generally has less activity during the Christmas period and the plan was that this time could be used for staff to become familiar with the new system and to deal with any implementation issues that arose. Indeed, the NOC decided to plan to further reduce activity through the Christmas – New Year period to give maximum opportunity for staff to support the go-live.
In July 2005, in agreement with NHS Connecting for Health, Fujitsu decided to change software suppliers from IDX to Cerner. Fujitsu’s contract with Cerner was signed in September 2005.
Although this left little time for Fujitsu, Cerner and NOC to prepare for the implementation of Cerner’s Millennium system at the NOC, the Trust decided to maintain the implementation date of 19 December 2005 for a variety of reasons:
• It had already taken steps to manage the demand on its services by reducing activity at its clinics and reducing the number of patients booked to clinics and requiring attention during the Christmas period.
• 400 staff had undertaken staff training, and a delay could mean a requirement to repeat that training.
• The NOC’s PAS was to reach capacity in February 2006.
• The Cerner system was well established in the United States and was in operation in Britain at the Newham and Homerton Trusts, and it was, therefore, regarded to be a ‘known’ product with an established record and fit to deploy.
• Any delay would have a negative impact on the remainder of the Southern Cluster roll out of the National Programme.
• Contingency plans were put in place by the NOC to mitigate against any disruption.
At that time NOC and its partners were confident that implementation issues could be addressed before a normal pattern of activity resumed in the New Year 2006. Maintaining the 19 December date for implementation gave no additional time for the project to take account of the introduction of an alternative system provider (the replacement of IDX with Cerner), but the supplier, the Trust and the Southern Cluster signed off the implementation timetable as the risks were considered to be manageable.
The Project Initiation Document for the introduction of the system at NOC was signed off by the Trust Board in September 2005. Three planning events were held by Fujitsu but these did not enable Trust staff to see a demonstration of the system which was to be implemented at the NOC. Cerner staff were not at that time available onsite to work with the staff to configure the system for the NOC. The Trust understood that the NOC would receive the system in operation at the Homerton, however, staff were not able to visit the Homerton to see the system being used. In this respect, as I cover below, one of the lessons learned for other deployments is the need to ensure adequate attention to data quality and data migration and that realistic on site testing and rehearsal of the new systems takes place prior to their introduction.
The problems which arose when the new system was introduced
A Radiology Information System – part of the Cerner system – was introduced on 19 December and has operated without major difficulties since then. A PACS system was later implemented as planned on 23 January and has been a very beneficial addition to the trust’s previous capability. This represents the first combined CRS / RIS / PACS go-live achieved.
On 20 December, the care record system went live. From the implementation date, the NOC experienced a number of problems. The main problems were:
• The process for initial user log on was much more complex and took much longer than expected. The impact of this was that the NOC’s original plan to log all users on to the system in two days was shelved, and key users only were logged on initially.
• The system did not start being used until 21-23 December and during that time a range of functionality and data migration issues came to light.
• On 22 December, the system went down completely due to a power failure at Fujitsu’s primary data centre, following which the in-built resilience for the power supply failed to operate as designed. In normal circumstances, this resilience capability would have been tested prior to go-live, but the compressed timetable for the NOC go-live meant that this testing had not been planned until January 2006. For the NOC the care record system and the radiology information systems were unavailable for a full working day. The NOC contingency arrangements were not designed for this scenario.
• The system reported that it was printing letters inviting patients to clinics, and yet it soon became clear that far fewer people were turning up to clinics than expected as they had not received any notification to do so. Conversely, other patients were turning up for clinics that they were not recorded as having been invited to. The impact of this was inconvenience to patients, wasting of doctor and staff time and a need to reschedule appointments. The missed appointments then resulted in a backlog of outpatient appointments building up. The NOC contingency arrangements could have been used as a preventative measure but this was not anticipated.
• There were differences between NOC’s working processes, in which clinicians do not participate in the appointments process, and the way that the system was designed to work in the United States, Newham and Homerton, where they do. This had the impact, for example, of making the booking of appointments to clinics very slow for administration staff operating a different process.
• Reporting – the system at the time of our visit was unable to generate performance reports and reports of activity in the way the Trust needs to be able to manage its activity effectively. As a result the Trust had been unable to report externally for three months (the Trust reports that this is now much improved).
• For the time being, the appointments process remains slow. This initially resulted in an increased backlog of patients awaiting appointments, as NOC had not easily been able to determine which patients should be prioritised, but the Trust reports that the backlogs have now been eliminated. A system change has been requested to cater for the fact that there are trusts where clinicians do not participate in the appointments process (a key difference to the way that the software is designed to be used).
In summary, the problems arose because:
• Insufficient time was allowed to prepare for the introduction of the new system.
• The Trust and supplier did not have time to ensure that all data was accurately transferred and that the specified number of tests on migrated data had been completed. One week before go live, data migration remained incomplete.
• Testing of the system and of its overall resilience was inadequate.
• Although NOC put in place contingency plans for possible failures in the system covering periods of 24 hours, 3 days, 7 days and one month, these plans were insufficient once problems arose, because they were not extensive enough to match the level and prolonged duration of the difficulties encountered.
• The reporting arrangements in the software were initially inadequate to meet the needs of the Trust.
• The Trust had not been able to benefit from lessons learned from other recent implementations.
The report of the Serious Untoward Incident (SUI)
It rapidly became apparent across the Trust after the go-live of the system that its introduction was causing the significant operational difficulties outlined above. Many of these issues did not appear to be easily and rapidly resolvable, and so the Trust undertook an operational risk assessment of these difficulties on 9 January. This identified high and extreme levels of risk to the running of the Trust which the Trust considered could potentially impact on patient safety.
Our enquiries confirmed that the NOC then prepared, on 12 January 2006, a Serious Untoward Incident (SUI) report to the Strategic Health Authority in relation to the introduction of the care records system at NOC. The NOC also emphasised to us, however, that it believes that patients have suffered no harm in relation to the incident it reported. The purpose of a SUI is to bring matters to the attention of senior managers so that problems can be managed appropriately and/or lessons learned by other organisations. An SUI can arise from any event which involves a patient, service user, member of the public, contractors, NHS staff or other providers of healthcare involved in the process of treatment, care or consultation on NHS premises and results or could have resulted in one or more of the following:
1) Serious injury
2) Unexpected death
3) Permanent harm
4) Significant public concern
5) Significant media concern
6) Significant disruption to health care services.
In this case, the SUI lodged related to items 4, 5, and 6 – that is, significant disruption to services and potential risk to patient and public confidence. The NOC foresaw likely disruption to service delivery for several weeks and months. It also reported a likely failure to meet all its national performance targets. It reported potential patient safety risks, but it is confident that, in the event, these have not materialised.
In addition, the Trust’s Audit Committee recorded concern that it appeared that patient records might have been “lost” in the system. It turns out that in fact this was not correct, and all records remained in the system following migration. However, prior to some of the system improvements outlined above, it was not possible to access the records using the reporting function, but as reporting has improved the records have become accessible again.
Action being taken to learn lessons from what happened during the introduction of the new system at the NOC
Fujitsu and Cerner have provided both high level and local support to the NOC to rectify the faults arising from this deployment.
In addition, all parties are seeking to learn lessons for the next deployments of the Cerner system within the NOC and within the Southern Cluster of the National Programme for IT. The principal lessons arising are:
• Not to go live with a new system until all data has been migrated correctly and is clean and complete, and that this is assured seven days before the decision to go live.
• Ensure that strong project control is applied to be sure that the introduction of a system does not proceed if critical tasks have not been completed.
• Ensure realistic testing and rehearsal take place for the introduction of new systems, and do not compress the testing time.
• Test and train for the future working processes that staff will be adopting once the new system comes into operation.
• Trusts need 10 to 12 weeks to train their staff on the new system to be implemented, and staff must be released for such training.
• Take sufficient forward information from old systems to allow for a three month contingency period, and that contingency plans are sufficiently extensive and robust to deal with prolonged disruptions.
The Trust, NHS Connecting for Health and the Southern Cluster Senior Responsible Owner explained to us that a full evaluation of lessons learned is being prepared by the Trust and its partners so that deployments elsewhere in the Southern Cluster can build on the experience gained through the implementation at NOC. Some of the planned deployments in the Southern Cluster have been put back to allow this process to take place. The Trust reports that this process is now largely completed.