Ten tips to consider

E-Health Insider asked a number of industry experts for their tips for achieving and maintaining high quality data as the NHS goes forward with its migration from locally-owned silos of information to shared systems under the National Programme for IT.

1. Executive sponsorship

An organisation's data is a valuable resource which should not be put at risk through an inadequate project management framework. In these financially constrained times it can be difficult to engage a trust board with any project proposal that involves spending money. Richard Zeronian, district sales manager in Emc's UK Healthcare division agrees that intangible benefits are no longer sufficient; efficiencies have to be the kind that show up on the balance sheet. He believes that there will be scope in most organisations to achieve significant savings, as well as improved performance, through rationalisation and /or virtualisation of their servers.

2. Understand your data

The higher the quality of an organisation's data, the easier it is to manage and the less risk there is of creating duplicates when systems merge. Theoretically, if it was of sufficient quality, trusts could migrate all of their data from their existing systems to CRS, resolving the problem of what to do with what is left behind.

The minimum they actually need to move across are all their currently active records, for which 95% of NHS numbers are expected to be complete. Many older systems have a history of different formats of NHS numbers, which may not have been cleared down. Locally used diagnosis and treatment codes will be meaningless when migrated to national systems. CfH have developed tools to assist trusts in improving data quality and their Back Office project is dedicated to the problem of duplicate records.

John Wiltshire, Stalis' sales and marketing director maintains that one cannot expect 100% clean data in legacy systems. Trusts might have been lulled into a false sense of security over the quality of their data because having a high percentage of NHS numbers is not as much of a guarantee of quality as they think. Some new systems will accept poor data because the NHS number is not a key field and not validated. Wiltshire adds that, even with the best tools, data cleansing requires a high degree of manual intervention and trusts should never underestimate the task.

"A key issue is being able to categorise all business data, and enabling users to search across the entire IT resource."

-- Keith Ricketts, marketing manager, Ardentia

3. Ensure business data can be categorised

Keith Ricketts, marketing manager at Ardentia said: “Data management means being able to use the data effectively when needed. So a key issue is being able to categorise all business data, whether in structured, relational databases and business systems, or in documents and the like, and enabling users to search across the entire IT resource. This is where enterprise search techniques come into their own.”

4. Account for changes in the law around confidential information and its accessibility

Data cleansing is not a one-off exercise for the purposes of CRS migration. Steve Tuck, chief strategy officer at Datanomic, emphasises that organisations should take this opportunity to review their validation rules and thus create defences against future sources of degradation.

5. Be aware of future scalability needs

Ian Anderson, HP's storage sales manager in the UK and Ireland, says confidential NHS data often needs to be stored for decades. Imaging modalities generate increasing numbers of files per patient and individual files are growing dramatically. Medical archives can triple in size in as little as two years.

6. Decide what proportion of data has to be on high cost, fast access storage, place the rest on low cost ATA drives

Trusts will always need access to a local data repository for ad hoc reporting, but there is concern that some LSP solutions will not offer that facility. Best practice currently is for data interrogations, especially large and complex queries, to be run against a locally-held copy of a recent data set. This is not to be confused with the need to maintain a searchable archive of that data which is not migrated. Says Steve Tuck, that archive must also be integrated so that it can be queried in conjunction with LSP data.

7. Emphasise the fundamental importance of data management

"If data migration goes well, the deployment goes well too."

-- Markus Bolton, director, System C

System C director, Markus Bolton, says data migration is absolutely fundamental to a smooth implementation but, from long experience of NHS projects, he believes too few organisations take this into account fully in their preparations. "If data migration goes well, the deployment goes well too," he says. Make sure you have allowed enough time in your plan to do the work and guard that time if the pressure is on to speed up the project.

8. Develop a testing programme

A testing programme should incorporate all trust activity across all specialties. Whatever the pressures on time and money, says John Wiltshire, this is an area where trusts should not be tempted to compromise.

9. Do not undervalue existing assets

Organisations could be forgiven for focusing entirely on new systems, says Phil Birchall, Intersystems' healthcare business development director, because that is how the national programme was constructed. Delays in delivering the new products mean that suppliers of existing systems are having to extend contracts and sweat existing assets. Having had a grandstand view of the English experience, the NHS in Wales is looking to build on current systems so that their future will be based on the interoperability of a diversity of systems.

10. Finally, says Steve Viner, HP's healthcare business manager for the UK and Ireland, do not overlook business continuity requirements - ensure that data is safe!