A tug of war over LHCREs is underway
- 12 February 2019
When the Local Health and Care Record Exemplars LHCREs were created, the official line is that they would support local direct care and enable population health management and analytics at regional level. But, as Digital Health editor Jon Hoeksma reports, tensions are emerging on whether they should also provide a bridge to a federated Care.data-like national data resource.
What is the core purpose of Local Health and Care Record Exemplars (LHCREs)? Are they to be designed to primarily support the direct delivery of care locally, together with regional population health management and analytics? Or are they a stepping stone to creating normalised data that can be linked together to create a valuable NHS national data asset and capability?
Officially, the line is that LHCREs should support both, starting with an emphasis on local direct care. But Digital Health News has learned NHS England is insisting the projects focus on providing normalised data for national use, potentially at the expense of the local care requirements. And tensions are arising as a result.
On 31 January, representatives from the eight LHCRE projects for an NHS England-organised Local Health and Care Record Programme Conference, at which Matt Hancok described LHCREs as the missing link to integrated care.
At a 5pm meeting at Elephant and Castle, immediately following the event, NHS England’s deputy CEO Matthew Swindells challenged the LHCRE leads on how they would standardise, or ‘normalise’, data to enable a common library of analysis tools to be run across LHCREs nationally.
“This is not what we signed up for”
One source told Digital Health News: “The bone of contention is NHS England’s insistence on persisting [storing] data at scale. And many of the LHCREs are simply saying this is not what we signed up for.”
They added: “Matthew Swindells had all eight SROs from LHCREs after the meeting and he made clear that this has to be about persisted data.”
The NHS England deputy CEO brings an in-depth knowledge of the US experience of population health management, having previously led on it for Cerner. He is said to be a leading advocate of the need to persist data at scale and developing a standard suite of tools that can run across each region.
“Matthew, because of his population health knowledge is anxious about persistence at scale,” another source told Digital Health News. “Although it’s not totally explicit in the LHCRE documentation it is strongly implied that you need persistence at scale.”
Without persisting and normalising (standardising) data at scale, it becomes very difficult to develop regional population health management. This is seen as key to the integrated care systems, into which 14 sustainability and transformation partnerships (STPs) have evolved and which it is hoped many more will become.
Developing a single data-layer at a regional level is also critical to the planned regional Digital Innovation Hubs. These are envisaged by NHS England as enabling researchers and academics to mine aggregated regional databases of anonymised NHS patient data. Three are to be developed from the five original LHCRE sites.
Taking a dip in the lake
In a data lake setup, patient data is ‘poured’ from local clinical source systems, and then blended together to create large normalised databases of anonymised patient data and enabling regional analysis.
One source linked to the programme said: “They [NHS England] can appear to have no interest in any part of LHCRE specifications other than data normalisation and sharing back to the centre. There is a tension between what people paid for and what they really want.”
The potential challenges here are two-fold. Firstly, not all of the LHCREs are yet designed to support direct care first and secondary uses second. Most of programmes, such as London, build on a patchwork of existing shared records initiatives. While many ‘persist’ data extracts for direct care coordination or some analytics and research, they have have not all been designed with this as their main purpose.
The need for open IG debates
Furthermore, a linked-up data lake approach runs into major information governance (IG) questions and begins to resemble a reboot of Care.data. This programme, which ran from 2012 to 2016, was axed after singularly failing to win the trust of the public or medical profession on IG.
Sources close to several of the LHCRE programmes say they signed up for creating regional data lakes of blended data to support patient care and analytics. But they emphasise they did not agree to then connecting these together to assemble a normalised national data layer built by linking the regional lakes, and doing so without obtaining local consent potentially risks losing public confidence.
A further source, a veteran of record sharing projects, said: “I think the objective can be firstly to support direct care, and also to provide access to context-rich sources of data for useful aims of service management and innovation, as well as appropriate research.”
However, they added: “We do not need nor support regional or national lakes of data, which would provide a focus for distrust from patients and people, be associated with complex issues of management of IG, and inevitable risk of data conflicts, and lack of clarity, and problems of data curation.”
One non-LHCRE CCIO working on shared records said there were strong arguments for creating blended regional databases. But they acknowledged that IG represented a significant potential stumbling block.
“There are a range of opinions. Personally, having seen some of the work done in the US by the large health exchanges, I can see the advantages from a population point of view of holding the data together. However, I completely agree that we need these IG debates in the open.”
NHS England declined to respond to detailed questions posed by Digital Health News but said in a statement: “NHS England is not creating a national data layer. The focus is on local health and care records.”
A spokesperson added: “The Local Health and Care Record programme will improve decision-support for clinicians by ensuring that clinically relevant information about a patient, from all parts of the healthcare system, is available at the point of care.
“This will help front-line staff improve care for patients, in London this approach has already seen success with patients accessing 111 services being identified as potentially frail, enabling proactive interventions in support of their care needs.”
15 Comments
Just two words … SCOPE CREEP!
The results of scope creep are almost always reduced quality, increased time, increased costs and reduced customer satisfaction.
The big question is “who is the customer here – Central Government or patients?”
Are NHS-Digital not already forming a huge normalised/linked healthcare record system:
https://digital.nhs.uk/data-and-information/data-insights-and-statistics/improving-our-data-processing-services
I’ve not heard much about the above lately.
The problem we have here is that there is no getting away from the need for a persistent ‘at scale’ data layer in order for patients (and this really is for patients) to be able to get the benefits in terms of having the best information available to everyone (and I include Clinicians, PHE, ops. managers and Commissioners in this group) to support their health and well being. However those of us with experience of building regional records know that this takes time (10 yrs plus) but it can work at a regional level, if we harness all the good practice across the country. In essence we need; an open debate and total transparency with the public, to make sure the our tech partners understand data is not part of their IP and finally to understand that if we focus on creating good quality ‘normalized’ data across a region then the NHSE/I will reap the benefits at a national level more quickly than trying to push their own agenda above everything else.
What are these major systems that can only produce a pdf?
Anyone who didn’t understand that the LHCREs were about persisting data at scale has not been paying attention. Maybe somewhat covert because of the nasty taste left by care.data but very definitely there.
We need to do population health and realise that benefits we can get from Big Data Analytics ML and DL (sometimes called AI)
We also need patient centred records that can properly support new model of integrated care.
BUT – We must not repeat the errors of care.data – No fancy footwork about “legal basis for use” , Section 251 or any other subterfuge. No pretence that rich patient records can be effectively anonymised. We need to let citizens decide how their data is used and we need to persuade them of the benefits of consenting to secondary uses.
When I look at the participation achieved by UK Biobank I don’t doubt that with the right approach patients will donate data for secondary uses. As Prof Joe McDonald says “The data’s there for the asking not the taking”
With colleagues from the Apperta Foundation I have outlined how we might start here apperta.org/coPHR
Yes I agree with what you have said, but as a Patient with an extensive PHR, I have no use case for putting my data into an NHS store.
It is already shared between apps using ‘banking style’ or ‘google’ authentication and authorisations.
If a clinician wants to view my PHR data, that’s possible and is not difficult. As a patient I can show them the data on my phone but also as a techie I can build a standard NHS API using FHIR and NHS logon as a hack in a day.
I can explain (and demo) further and will try to at the Rewired Hack.
Im not convinced by arguments we don’t need a persistent cleansed semantically harmonised record.
Yes we need systems with standard open APIs and this will accelerate innovation. However, it has the potential to create a hairball of interactions with little control, poor governance, difficulty swapping systems, poor population health, poor early intervention and prevention, poor data quality, poor performance and an inability to roll back to a past information state.
A persistent record in a hub a spoke model enables all of these and having just spent a week at HImss, it’s the way most of the USA is going. NY already has one with over 17M records and is seeing the advantages.
If you’re persisting, you need standardised algorithms across LHCREs. Eg, Are we defining diabetics in the same way, whether by diagnosis, tracking of HB1Ac or the presence of insulin medication.
Making sure we design for the future implementation of this technology is practical, makes sense and doesn’t detract from delivery of local direct care. Quite the opposite. It enhances it.
Well done NHSE. LHCREs are the correct way to go!
Beautifully put Ewan. We have spent 4 years having an open conversation with the citizens of the North East about data sharing and have had considerable success with it by applying NHS Digitals Design Principles https://beta.nhs.uk/service-manual/design-principles including the need for openness and transparency, if applied to LHCRE we can avoid Care.Data2 and achieve the benefits we all want. A culture of secrecy won’t help.
What are you saying Paul. I’m confused by your post.
Do you mean “Im not convinced by arguments *that* we don’t need a persistent cleansed semantically harmonised record”
Or
“Im not convinced by arguments. We don’t need a persistent cleansed semantically harmonised record.
IF the former I’m with you if the later I’m nit.
As for learning anything from US healthcare I’d always advise caution . Let’s remember their health system is the most broken and expensive in the developed world. Let’s take any good ideas they have but remember their IT is designed for a very different environment that we would not wish to emulate.
Quite typical of NHS England to start to move the goal posts after the match has started and as usual be unclear in specifications – Its about time they (NHS England and Government) started to realise it is them that have held back national NHS IT systems for years with their dithering about…
@rozman, 100% agree, you are bang on. The interoperability challenge is much bigger than people appreciate, evolving standards are of little help when some of the source systems can, at best, produce a pdf. Whilst the vision for population health on a national level is commendable, we have to crawl before we can walk, before we can run.
Whilst data normalisation should be the long term goal, with many systems currently only able to produce a PDF its not going to be possible at scale any time soon. Common sense says your first objective should be to support direct patient care with a journey to normalisation and pop health.
Getting systems and people sharing is hard enough as those on the ground trying to deliver it will testify, not sure the Centre understand the reality of the ask.
IG is only a headache because these LHCRE vendors, or the providers they sell to, think that the data belongs to them. They implement closed systems to protect what little IP they actually have (it’s a patient record, not a mathematical algorithm!) and while they integrate, they do their best to avoid it…
There has to be a systematic change in how these vendors and providers see the data.
I’ve seen how little understanding there is in the market, and how vendors are driven by the quick sales turnaround, with a view to dump their crappy software onto the next big fish that acquires them.
They don’t care about an IG-centred product that adds real value to the industry. They’re just after a quick buck.
@rozman, 100% agree, you are bang on. The interoperability challenge is much bigger than people appreciate, evolving standards are of little help when some of the source systems can, at best, produce a pdf. Whilst the vision for population health on a national level is commendable, we have to crawl before we can walk, before we can run.
Excellent article.
The centre should not be advocating legacy architecture, experience from trusts, councils and other UK countries (Scotland) have already shown far more modern ways of cooperation + sharing which have patient focused security/privacy built in (and increase them as this centralisation will do).
Lets not spend loads of money moving data to a central database. Instead open up systems via standard API’s and enable coordination between clinicians with standardised messages – it’s technically cheaper and easier to do, improves patient care, protects patient information…..
Comments are closed.