Data remember
- 17 May 2005
Daloni Carlisle
The thought of spending hours, days or even weeks checking, rechecking and finally inputting data that’s gone missing during a botched migration is enough to make any GP think twice about changing their system.
But with NHS Connecting for Health pushing for practices to move to systems hosted from central data-centres, it’s a choice that many more practices will have to think through.
Ewan Davis, chairman of the British Computer Society’s Primary Healthcare Specialist Group, has some simple advice: "Do not subject yourself to a major system upgrade that’s not going to take you anywhere that’s significantly different to where you are today."
Data migration, he says, is a critical issue and one that has so far had relatively little attention. "We have all heard the horror stories of data migrations that have been poorly handled," he says. At risk is not just patient safety but also practice income tied up in the Quality and Outcomes Framework.
Horror stories
"Problems with data transfer are well nigh intractable"
— Dr Mary Hawking, GP, Dunstable |
Mary Hawking, a GP in Dunstable, became the most famous of these horror stories after publishing her experiences in the Journal of Informatics in Primary Care in 1995.
She wrote: "When we turned the computer on the [morning after a data transfer], it was immediately apparent that something very strange had happened to our data. For example, ‘knee joint pain’ had been changed to ‘arthralgia of the lower leg’, and ‘plantar fasciitis’ to ‘plantar fascial fibroadenosis’."
She says nothing much has changed in the intervening decade. "Problems with data transfer are well nigh intractable and come basically because data bases are constructed differently and everybody uses them differently."
Done well, data transfer takes an enormous amount of work on the part of practice to clean their data prior to extraction and good support from your supplier as its formatted and imported into the new system, she says.
Done badly, and you can end up with data in the wrong place and free text attached to the wrong diagnosis and no means of knowing why a diagnosis was made in the first place, she adds.
A recent exchange of experiences at a practice managers’ email forum confirms this. Answering a question from a new practice manger about to embark on a system change, they variously warned her to keep paper records and be prepared to work with those for weeks after the data transfer.
One wrote: "The biggest issue as you are probably already aware of is the data transfer. You can only hope it goes as smoothly as possible, but there will undoubtedly be problems of some description, just keep your finger crossed that they are few."
Yet it potentially affects all practices, not just as they update or amalgamate but also as they come under pressure to move data off practice-based servers and into centrally-hosted local service provider data centres.
Not all-or-nothing
Right now it’s not a clear-cut choice. The official message from the centre is that GPs would be wise to move to an LSP-hosted system to make best advantage of the national systems such as Choose and Book.
That’s being countered by the British Medical Association’s General Practitioner Committee and Royal College of GP’s joint IT Committee which issued a statement in 2004 saying that "moves to new systems would not be contemplated until data migration issues had been resolved". So far, no such resolution has been forthcoming.
Mr Davis says it’s not an all-or-nothing choice – yet. "QMAS was delivered without moving anybody into a data centre and the pilots for choose and book are with practices that are not using LSP-hosted systems," he says.
"From a technical perspective it’s entirely clear that the programme can deliver in either environment. It’s more that the data-centre hosted solution is the direction of travel that the national programme wishes to encourage."
That’s because of the next stage of development envisaged – the national shared record that will almost certainly require GPs to migrate their data into a data centre.
So, as Mr Davis points out, moving now to an LSP-hosted system will not prevent the need for another migration in two to three years time as Carecast and Lorenzo develop their primary care functionality.
But there are practices facing a pressing need to upgrade their system right now – for example an out-of-date system no longer supported by the supplier or one whose supplier is not compliant with the national programme.
Here there is a considerable experience to draw from – although precious little in the way of best practice guidelines from the professional bodies.
Responsibilites
"One burning issue is to do with data quality and data checking. It has to be done by clinicians. But the fact is that clinicians are usually busy seeing patients"
— Simon Lovegrove, director of GPData |
There are, say the experts, four basic steps to successful data migration. They are:
- Cleaning
- Extracting
- Formatting
- Loading
Cleaning the data, along with extracting it, is an NHS responsibility. In data centre-hosted systems the last two are the responsibility of the LSP.
Simon Lovegrove, director of GPData, says: “Only a GP is qualified to judge the success of their own transfer but there are several key issues.” Among them is the recognition that the task is complex and that each practice is unique.
He adds: "One burning issue is to do with data quality and data checking. It has to be done by clinicians. But the fact is that clinicians are usually busy seeing patients. They need support and funds to find the time to check their data."
This was a point picked up by a report on data migration commissioned by NPfIT and leaked to EHI earlier this year. The level of clinical involvement was likely to be the rate-limiting factor in migrating practices into the new LSP-hosted systems, it said.
The other issue is clinicians’ expectations – a point highlighted by both Mr Davis and Mr Lovegrove. "There are two key elements to getting this right," says Mr Davis. "Converting your data successfully is the first. The second is the re-engineering and retraining that comes as a result of the change of system."
Mr Lovegrove agrees. "We can transfer the data but it cannot be assumed that all functionality of the existing clinical system will be preserved. Practices have to face up to that."
As one practice manager puts it: "A system change has a huge effect on your practice with months of preparation before hand, huge disturbance to your working efficiency for months afterwards. Only slowly do the gains come through. Many of the things you held sacred from the old system just may not be possible with the new system."
Mr Davis advice bears repeating: data transfer is a risky business and only worth it for a very clear gain.