How up-to-date patient data can help the NHS

NHS systems are only as good as the data they rely on. Outdated or fragmented data may not only lead to trusts and integrated care boards not getting the best value for money from their IT spend, but these problems may become further embedded as data gets spread across a system and migrated into new software and electronic patient records.

May Rowe-Spencer, principal population health analyst at Lewisham and Greenwich Trust, is one of the fortunate data analysts lucky enough to have access to live data feeds from acute trusts, primary care and community services, making inconsistencies easier to spot.

“Because we get live feeds in, we can see the mismatches between what different organisations hold,” she said. “That has meant that we’ve done quite a lot of work… taking demographic data like ethnicity and language, and also address, and so on, from the GP record and seeing where there are mismatches in the trust and see if we can update the records. An issue that poses, of course, though, is that that’s a very manual process at the moment.”

This system can ensure patients’ GPs are aware of any pending operations so any health conditions such as diabetes or hypertension can be optimised beforehand, lowering the risk of the operation being cancelled, she added. Having an up-to-date patient tracking list is vital for this.

Ms Rowe-Spencer pointed out it was much easier to correct details at source where the information was “right in front of you”, adding: “The further removed you get, the harder it is to make those decisions about whether something is right or wrong.”

Interoperability may pose challenges to sharing data, but even having the same system does not guarantee data is comparable.

ThinkTribal principal data consultant Richard Sutcliffe said: “Even if you’ve got the same system elsewhere, you don’t necessarily have the same standards and same business processes in place.”

The focus is often on top-level national reporting, he said, which could mean the numbers were regarded as important, but the context of the information underneath got overlooked.

“A good example is probably community [services] where you can easily see the number of visits that have been taken but not necessarily see why those visits were taken and what was done during those visits,” he said. “It is important to try and have a complete picture of your data, or have as complete as you can have, so you can make sense of what the data is telling you.”

Hundreds of thousands of data issues

These are the sorts of problems Experian Data Quality often comes across when dealing with clients, said senior account manager Rachel Tavener, but some of the most troublesome areas are with simple patient data.

“It’s making sure that when somebody is presenting themselves for the first time at a hospital, to a GP etc, that the data — just simple name and address — is recorded correctly in the system,” she said. “I know that sounds really simple, but something without an error and in the right format means that the initial letter goes to the patient at their address. There’s no delay in the treatment from that point of view.”

Around 60 per cent of trusts use Experian’s address validation system, she added.

Migration to a new system is sometimes when data quality issues become starker with simple things like formatting issues making the process much harder. Having ongoing monitoring, which picks up data quality issues, can help to avoid this, but Experian also has a data platform that can be used on-site to ensure that, when data is transferred, it is ready to use on day one, Ms Tavener added. To give some idea of the scale of the challenge, she noted Experian had worked with an ambulance trust, which found 400,000 data errors or data that needed investigation.

But what should NHS organisations having problems with data do? For Mr Sutcliffe, the important thing is making sure there is a strategy in place for any migration or linking of systems.

“Data cleansing is probably the first thing that you want to be doing in that sort of position… if you’ve got bad quality data in the first place, it’s going to make linking those data sources much more difficult,” he said.

“It’s really looking at what data you have currently, how you can try and make that consistent, accurate, timely, as complete as possible, and then looking at how you can then apply the mappings across those.”

Often the issues encountered were differences in processes and ways of working between organisations and departments, Mr Sutcliffe said. When systems were migrated or linked, these issues needed to be addressed as well as the quality of the data. “Otherwise, even if you fix the problem up front, it’s just going to continue afterwards and potentially get worse,” he added.

It might not be possible to deal with all the data at the same time, so trusts would need to prioritise, he said, and look for their “golden sources” — the data they trusted the most.

However, there can be inconsistencies even within hospitals. For example, Ms Rowe-Spencer said sometimes the same patient had different results, which could mean talking to the people involved to find out which data they would rely on most. “It requires a lot of talking to people, and then requires building rules into your systems to prioritise the different fields that you know are more reliable,” she said.

And buy-in at board level was important, stressed Ms Tavener, adding scrutiny, testing, and oversight are all needed, but improving data governance could ultimately save time and resources.

An on-demand version of this webinar is available.

To access the recording, visit this web page and click play.

If you have previously registered for the event, clicking the link will immediately bring you to the recording. Simply press play.

If you have not previously registered, complete the form that appears. You’ll then be able to immediately watch the recording.