London Borough of Camden senior business analyst Stuart Farina has worked at polling stations for many years, and met plenty of frustrated citizens who discover too late that they can't vote.
“People turn up to vote, they’re not on the register and you tell them,” he explains. “And they say, ‘But I pay my council tax’ – and you say, ‘The two things aren’t linked-up’.”
Some foreign nationals have to put up with taxation without representation, but British, Irish and Commonwealth citizens – as well as for now EU nationals in all but general elections – who are resident in the UK can vote. So Camden invites everyone paying council tax but not on the electoral register to join it.
This is possible because the council has joined 16 different sources of data on its residents through a master data management (MDM) system. It has not replaced existing databases for areas including council tax, social care, housing benefit and the electoral roll, but uses MDM to link the records – essentially doing what the voteless locals assumed it did.
As well as inviting people to join the electoral register, it also allows it to automatically process 78% of free transport pass renewals for older and disabled people without them having to complete any paperwork.
The borough went live with an IBM MDM system in April 2012, having decided against phasing in data sources four at a time.
“Where we had blockages in getting data out of old legacy systems that were a bit clunky, we just concentrated on another dataset,” says Farina.
He adds that if users had tried the system with just some of the data, they might have dismissed it: “I’ve worked on projects before where it’s been done in that phased approach, and users can just lose interest in it.”
Camden has found many uses for the system, with more than 300 approved staff in 35 teams having access. The first type of work was data intelligence, which allows the council to produce reports from multiple parts of the organisation with greater ease, such as a government social care return that includes information on education.
Making fuller use of its own data – such as on numbers of infants for education planning – allowed it to end a £13,000 annual contract with Mosaic, a geographically based classification system run by Experian, and halve its spending with credit reference agencies, saving a further £10,000 a year.
It also saves residents time. “One of the ways we used to validate residency was to ask people to send in their council tax bill,” says Farina.
“We’d sent it out them, then we were asking for a copy back to prove residency. Now we have access to the data in the MDM, we can do that without having to get the resident to send in information that we sent them or they have already provided.”
Camden has also used the MDM to tackle fraud, including tracking illegal subletting of council houses, where each misused property costs the council £18,000 a year. Farina says the council has identified 12 properties that have been illegally sublet, saving more than £200,000 annually.
The council has extended this work to school admissions, to check a child lives within a school’s catchment area. It allows staff to prioritise checks by looking for recent address changes: some children move in with grandparents or parents rent flats opposite a school to apply.
“If we know through the council tax register that someone’s been paying council tax at a property for 12 years and they haven’t just moved in, we can let them go through,” says Farina.
In the most recent admissions round, Camden withdrew five offers of school places after detecting fraud, more than all of its neighbouring boroughs put together. Staff can realistically spot fraud at the point of application, well before children start at a school, rather than disrupting their education afterwards.
Finally, the MDM helps improve data quality by highlighting potential duplicate and split records, which is particularly useful in social care – if officers only have part of a client’s notes they may lose the chance to prevent a dangerous situation developing. The children’s services team check every new referral on the system, with the aim of giving themselves the most comprehensive set of information on the child.
MDM to join up NHS and local authorities
Some areas are linking across organisations, using MDM to join health records held by NHS organisations with social care data held by local authorities. This can be aided through the UK’s identification numbers for health and social care, including the NHS number in England and Wales and the Community Health Index in Scotland.
But these numbers are not a panacea, according to Dan Cidon, chief technology officer of NextGate, which carries out data matching based on statistical analysis of patients’ demographic information.
“The percentage is not close to 100% in terms of the records that have an NHS number on file, and that percentage goes down substantially when you start to look at things outside healthcare,” he says.
In social care, use of the NHS number was only mandated in October 2015, and the company has integrated some social care systems that have no use of the NHS number at all.
NextGate has found the same NHS number used for two people as well as one person using two NHS numbers.
“In healthcare alone, there are a lot of discrepancies, sufficient that people recognise that the NHS number by itself is not enough to ensure that you have that 99%-plus coverage of your patient population,” says Cidon.
Scotland’s Community Health Index works better, he adds, where the company has worked with NHS Dumfries and Galloway to integrate data on the area’s children.
“It’s a smaller population, it is a more tightly controlled process for issuing those numbers and therefore the duplication rate and the mess is far lower in Scotland,” says Cidon.
Northern Ireland has two advantages: a carefully-managed health and care number system introduced in 2004, and integrated health and social care trusts. It has joined up records across all its six trusts through the Electronic Care Record project, implemented by Orion Health, and takes a robust approach to maintaining data quality.
The one million daily incoming messages to the system are checked against the health and care number index, by surname, date of birth and gender.
“If any of those mis-match, we exclude the message from the database, because we can’t afford to tell a doctor we’re using probabilistic matching and that we’re only a certain percentage confident that this patient record is the one you want,” says Gary Loughran, head of the eHealth programme at Northern Ireland’s Business Services Organisation.
Across all systems the match rate is above 97%, rising to 99% for labs systems. “It leads us to believe that in 12 years we’ve matured and mastered our own data,” says Loughran.
The system also locks records from trusts when key data does match what is held on the numbering index. This means clinical users have an incentive to sort out problems.
“They would demand of their information governance group: ‘Why is this record locked, can you please help me unlock it by improving the data quality?’. That had the single biggest impact in recent years on our data quality,” says Loughran.
The project has also benefitted from the fact that Northern Ireland had already centralised some of its health and social care IT such as through regional data centres, he adds: “We came from a position of relative strength in comparison with a lot of people. You can’t sprinkle magic dust over your IT and systems and hope to get an electronic care record.”
Some areas of England are applying MDM to health and social care, generally at city-scale.
The Leeds Care Record is used by the city council, the city’s acute, mental health and community NHS trusts and its three clinical commissioning groups to share patient data. It was developed from Leeds Teaching Hospitals NHS Trust’s in-house patient pathway manager clinical portal.
Tony Shannon, who previously worked as Leeds Teaching Hospitals’ chief clinical information officer, says there are advantages to a project being clinically led across a city.
“It took about a year to agree the information-sharing agreements and the governance,” he says. “It was because of those clinically led and hospital-led factors that I think it happened.”
People liked the fact it was a system from the local hospital, believes Shannon; at one GP surgery in Leeds 200 people opted out of Care.data, the since-cancelled England-wide data-sharing project, but only two refused to join the Leeds Care Record.
“The complexity of healthcare and health IT means it’s always best to start these initiatives relatively small then build out,” he says. There are discussions about a shared record system that works across Yorkshire, but this would build on the existing Leeds work.”
As an alternative way to share its work, with funding from NHS England, Leeds is offering an open source version of the software known as Ripple; in June 2016 it appointed Lockheed Martin now Leidos to help develop the project along with SMEs, including Answer Digital Health.
It is being trialled in South Tyneside and there are discussions on using it in Somerset, for Leeds’ person-held records and by a hospital in Dublin. In most cases the NHS number will link records, while in the Republic of Ireland the new Individual Health Identifier will perform this role.
“The Leeds Care Record does a very good job in Leeds, but by the nature of the architecture it’s relatively bespoke for Leeds,” says Shannon, who is director of Ripple.
The project uses a building block architecture where elements can be customised without breaking links to other parts: “You have to allow for local flexibility and customisation,” says Shannon.