UHBNFT shares NPfIT lessons

The University Hospital Birmingham NHS Foundation Trust was an early adopter of the National Programme for IT (NPfIT) in a teaching hospital trust. Andrew Haw, director of ICT at the trust and chair of ASSIST, describes some aspects of the implementation project, paying particular regard to reusable lessons for others. 

University Hospital Birmingham NHS Foundation Trust (UHBNFT) is a large acute teaching hospital trust with 6,600 staff, in the North West Midlands Cluster. The Cluster covers a population of over 12 million and 149 organizations. It was the first large acute trust to take NPfIT services, and converted to the Local Service Provider (LSP)'s reference solution for Patient Administration System (PAS) in January 2006. The LSP is CSC and the project began in April 2004. 

One of the key aspects of any systems implementation is how well the business and clinical processes and the new software fit together. UHB was expecting to, and did take advantage of, the new software to change processes and to force better adherence to good practice in patient management.  

Sharing trainers made sense 

One of the other interesting developments has been trusts and PCTs working together in local health economies. Using a common model to calculate implementation effort, the five South Birmingham trusts were able to show that their collective implementation plans fit together in such a way that total collective IT training resource was constant over a three-year period. In other words, it made sense to share trainers.

A South Birmingham IT Training Agency was created, with a budgeted headcount of 16 trainers. This has worked well as planned.

The NHS has to find enough trainers who can become product experts in their own right. The recruitment of trainers and the availability of training rooms was a logistical hurdle. The Trust and Agency had to take 14 trainers through extensive product training, nine of which had to be recruited from outside the Trust.

A separate building had to be leased and furnished for training. The effort required to construct end user training schedules should not be underestimated and communication with all line managers is imperative if the training is to be a success. 

It may have been thought that any trust PAS could be easily replaced without too much effort, once the product was sufficiently standardized. At UHBNFT however no fewer than 36 clinical information systems were interfaced to the old PAS and a number of interfaces were bi-directional.

Not only was a new interface engine required to handle this but all interfaces had to be re-written and thoroughly tested before go live. Some 12,000 messages are now received daily from the CSCA Data Centre. 

Data migration 

One of the debates in the Trust was the value of historical data. When paper health records files are kept for 10 years after the patient has last been seen, it is important to provide a route to those notes should the patient return. However, the effort involved in finding NHS numbers and in tracking down suspected duplicate registrations needs to be justified. 

The Trust's analysis showed that there was an age of data beyond which the benefits of retention begin to be swamped by the cost of loading aged data, together with the reference data that was current at the time that the operational data were created. Accordingly the Master Patient Index of all patients attending in the last 10 years was converted in entirety, whereas only 3.75 years worth of activity data was converted. 

The Go Live period 

The trust commissioned its external auditors, KMPG, to perform a project readiness review report and this was part of the go live decision.

The legacy IRC PAS was put into view only mode at 6pm on Thursday 26 January and the data extracts commenced. All users had been made aware of a simple downtime data capture system, linked to a home written 'mini PAS' application on the Intranet, which provided the means of capturing inpatient activity data over the expected 84 hours of downtime.

Over the weekend a total of seven checkpoint meetings were held with trust executive directors, mostly by conference calls. Early on Sunday morning the system was prepared for operational use and the Spine access was switched on.

From Sunday morning the control centre was established so that PAS Support, EPR Team, IT trainers and CSCA could be co-located. CSCA support resources (Early Life Support) were also placed on the IT service desk to pick up on calls that need escalation to CSCA. 

From 8am Sunday the team began validating the data in the application and in the general extracts, which the Trust requires on a daily basis to construct our own information reports. From noon the team began capturing all of the data from clinics and wards from the period from Thursday night to Sunday afternoon, and entering these into the system.

At 4pm we had obtained good results against the data migration tolerances that were set in advance. Only follow-up outpatient appointments (610 rejected out of 71,964 expected) were above the data quality thresholds agreed by the project board.

The configuration of the clinics concerned was fixed during the day and the appointments resubmitted to cure this problem. This was confirmed with executives at the 6pm checkpoint, which effectively confirmed the safety of going live at 7am the following morning. 

The go live effort was a genuine team effort with CSCA, the project team, health informatics and the data take on team all working together effectively. 

Operations in first two weeks 

For near user support, we had requested additional product specialists from CSCA to supplement our own trainers and IT Agency trainers in wards, clinics and secretarial offices. Collectively these were known as floor walkers, and up to 34 were deployed in red t-shirts between 8am and 2am in operational areas (the majority from 8am to 5pm).

The floor walkers were in radio control with the control centre. The training manager could direct floor walkers to users with urgent problems as the calls were received on the IT service desk. LanDesk was used by all staff in the Control centre to control a user's screen and mouse while the user was on the phone to provide the relevant help and support.

On the first day, there were no application issues, just a few oddities in the way our data had migrated. We had 150 Lorenzo and PAS related service desk calls compared with normally 50 (see below), and no increase in PC related calls. 

Calls about smart cards and locked cards were lower than expected. Pleasingly 34 of 36 interfaces to departmental systems were working by the end of the day. Some issues have arisen due to the way in which IRC PAS was used. 

It was not uncommon to see 'To Come In' transactions that were quite aged with no outcome recorded (no attendance, no DNA and no new offer of a TCI). Due to the fact that UHB shared, and continues to share an MPI with another local trust, there were some special measures to handle the migration of data because the trusts have patients in common.

This is the first time two acute trusts have shared an MPI and some compromises had to made to load all the data: there are 80,000 patients in common. The issue arises when a patient has two MPI records, one of which does not have an NHS number; there are 17,000 of these. 

Support calls 

For the first four days, total service desk calls were 42-58% above normal.  From day 7, this had fallen to 7-8% above normal.

The table below summarizes the first eight working days.

Mon Tues Wed Thurs Fri Mon Tues Wed
Total incidents logged on SD 353 343 310 312 231 291  257 234
Percentage above normal 58 44 42 56 36 30 8 7
Total PAS and Lorenzo incidents  65 124 102 116 63 85 75 69
Percentage of calls to tape 60 61 44 21 25 17 20 20

A relatively low proportion of these calls were transferred to the national service desk.

In conclusion, the implementation went as well as it could do, considering:

  • the way in which IRC PAS allowed users to use it in the past, and the condition of the data migrated.
  • the changes in processes that users have to come to terms with.
  • that we are the first acute trust to tackle interfacing and batch printing in the volumes currently experienced, and to share an MPI with another.

The system is also demonstrably far better at policing adherence to best practice and detecting when a user, clinic, or speciality has deviated from good practice.

Andrew Haw: andrew.haw@uhb.nhs.uk 

More Articles