EraserDaloni Carlisle 

It’s all too easy to make a ‘tpying errer’. When it’s E-Health Insider it just makes us look sloppy. But when it’s a GP practice it can determine whether a doctor has access to a new patient’s notes.

An exercise to test the quality of GP lists and how well practices cope with changes of address and deaths has uncovered just how easy it is for spelling mistakes to slip through the net.

The National Duplicate Registration Initiative (NDRI), found 4,781 misspelt Micheals (compared to 799,565 correct Michaels), making them difficult to track should they move.

The NDRI also found 135,000 patients on GP lists who were, in fact, dead. Over 10,500 of these had died before 1990 and so had been on a GP list for at least 15 years before they were removed. The longest case was a patient who died in 1964 but was not removed from the GP list until 2004.

These are some of the several fascinating results of this summer’s final report of the NDRI, a massive data matching exercise which started in 2003/04 and was undertaken by the Audit Commission as part of the statutory audit of primary care trust and local health board accounts.

In brief, it involved extracting electronic details of the 56 million patients registered with GPs in England and Wales, undertaking data matching with records from, for example, the Department for Work and Pensions and the Home Office then feeding back the matches for review.

“We also used fuzzy logic that would pick up things like the misspellings,” said project manager Darren Shillington. “That’s how we picked up the Micheals as well as more unusual one-off mistakes.”

In total almost 1.5 million matches were fed back to the 87 National Health Applications and Infrastructure Services (NHAIS) sites who manage the patient list data for all PCTs and LHBs. They in turn worked with their local GP surgeries to cleanse their lists of any duplicates.

Cancelled registrations

The Audit Commission’s main concern, as one would expect, was value for money. The audit took place as the new General Medical Services Contract came into play. With payments based in part on patient numbers, the Audit Commission was keen to make sure practices had accurate records.

By and large they did. The NDRI has led to 185,000 registrations being cancelled, saving £9.5 million. A significant number, to be sure, but representing only 0.3 per cent of the population.

Or as Mr Shillington puts it: “From our perspective we were looking for audit assurance. Is the data reliable and materially accurate? We got considerable assurance about the accuracy of payments being made on the basis of the data. Yes there are anomalies but we found that the system is solid.”

The NDRI highlighted the difference between NHAIS sites responding to the exercise. Each participant received a CD-ROM with matches that needed examining and while the Audit Commission provided guidance, each responded in its own way.

For example, Essex NHAIS sites automated a validation of 4,242 deceased persons’ matches by using Microsoft Access to compare all persons who had been removed as deceased since the NDRI data download date with the matches derived from the NDRI. Details of any persons not removed were fed back to the relevant practices, which were given two months to identify any patients who appeared to be alive.

While Essex was able to automate, others resorted to manual methods for this task. Some are still at it.

NPfIT and data quality

Mr Shillington said variation between NHAIS sites was to be expected. “Populations also vary,” he said. “Some sites have a very transient population and therefore a more complex task.”

The NDRI did not take place in isolation, however. It was timed to have the maximum benefit for the National Programme for IT and the Information Quality Assurance Programme (IQAP), both of which have an interest in data quality, not least because the basic demographic data for the proposed national care record will come from GP lists.

It identified several lessons for this wider effort. For example, the system allows duplicate registrations for the same NHS number – an issue that the report says must be addressed in the new patient demographic system. Temporary NHS numbers are still in use; these will not be allowed in the new system.

The Audit Commission has fed these lessons back to NHS Connecting for Health and IQAP. But for some that’s not enough.

Paul Cundy, chair of the BMA and RCGP’s joint IT committee says: ‘This is not a new issue. GPs have been ghost busting for years. But it highlights precisely why you should not have a single record for everybody.”

An out of hours service or A&E department will not be able to find any “Micheals” or separate the Lawrences from the Laurances, he says. “That’s why we would prefer a system where the emphasis is on information beamed from place to place and matched by the person.”


The National Duplicate Registration Initiative