Implementing Laboratory Medicine (Pathology) Standards
Laboratory Medicine (or Pathology) in the UK has been a challenging area for standards for a long time. When I took on the role of leading the development of future pathology standards for England (via NHS Digital), I was offered many sympathetic reactions – since many had tried and failed. However, this post is about the future not the past – the way to get the new Pathology standards based on #SNOMED CT, HL7 FHIR and UCUM implemented/adopted in the lab systems d. If you didn’t know, in England new (#NHSD, #NHSX) Pathology Standard(s) were approved earlier this year. So this article has a slight focus on the UK, but could be of interest more widely too, even if you don’t use SNOMED CT in lab medicine or use SNOMED CT in a non-lab medicine context.
Laboratory Medicine (Pathology) is challenging in the UK and elsewhere for broadly similar reasons. At the risk of oversimplification (merely to present a story), here are some important reasons. Apologies in advance for anyone who might be offended by my personal views.
- Lack of investment in Pathology/Laboratory Information Systems (LIMS)
- Fatigue of too many promised and failed saviours (standards)
- Disconnect between `future` and the `day to day` operational needs
Lack of investment in LIMS
This has been long acknowledged in the Pathology field. Many systems now in place are a decade old, if not older. There are still black screen systems (Unix boxes) that computer geeks/nerds/hacker types would recognise from the Matrix movie (side note, Matrix movie itself is 20 years old)!! So there is no doubt that these systems need to be upgraded. This is of course part of the bigger NHSE Pathology Consolidation work, which is also looking at upgrading the infrastructure.
The lab/pathology professional community have been promised on several occasions that a shiny, new `saviour` project/programme/standard would solve the issue. The professionals believed the promises that PBCL, NMLC, LOINC, etc were the thing to believe/invest in. Unfortunately, over a decade of this has left the professional community a little wary (and rightly so). Let me say this straight up – I am yet to meet a representative from the pathology professional community who does not believe strongly in the need to standardise. So it is amazing that we managed to create a `yet another standards initiative` reaction in such a passionate community!
Disconnect with operational needs
We `standards people` need to refocus on operational needs where standards are designed to be used.
- Standards only address/solve a tiny fraction of the operational needs/problems.
- Interoperability is cool and don’t forget humans use systems. So not all clinicians and patients want to see long strings like :
- `Mass concentration of methylenedioxypyrovalerone (bath salts) in urine (observable entity)`
- `Substance concentration ratio of 5-aminoimidazole-4-carboxamide ribonucleotide to creatinine in serum (observable entity)`
- `Arbitrary concentration of goosefoot, lambs quarters (Chenopodium album) pollen specific immunoglobulin E antibody in serum (observable entity)`
- A lot of clinical practice depends on workflows that add value, helping over-stretched healthcare professionals deal with their work more efficiently. So a `new standard` that breaks existing clinical workflows is not going to be welcomed or supported on the ground
- Standards don’t always stay in sync with evolving needs. When Covid-19 hit us, it was being recorded/reported on before standards were brought up to speed.
None of the above are criticisms of standards or imply that those who work in data standards don’t recognise these issues. So please don’t jump on this article defending standards… 🙈
Implementing/Adoption of Laboratory Medicine standards
Scotland is in the process of migrating to a single LIMS and wants their programme to adopt standards from day one. 14 boards (regions) need to be migrated to a single LIMS (don’t know if it is one or more instances). In listing multiple important issues above, I am trying to highlight that addressing only one is not going to work well. To ensure adoption/implementation of standards, we need to address all areas together – so think:
- New LIMS system procurement with standardisation in mind
- Engagement with pathology/lab-medicine professionals to ensure buy-in and ongoing governance (more later)
- Ensuring operational workflows/needs remaining unaffected (if not improved) by new standards
Let me complicate the picture a bit more by saying that for over a decade a lot of LIMS have existed as `data silos`, not merely within hospitals, but also possibly across disciplines (e.g. blood sciences vs microbiology vs histopathology).
Anyone with the task to bring such data silos together as `sanitised/harmonised` data using standards knows the mess there is to deal with. This was my experience when we tried to bring data from multiple hospitals in Singapore into their national EHR (NEHR). So anyone who thinks – we’ll just give busy clinicians/professionals access to a SNOMED CT browser and they’ll sort out their migration/mappings/refsets and somehow magically everyone will identify the same concept/lab-test/code has never experienced clinicians/professionals disagreeing about what a name/thing means… 🙈 You always need a system of review/discussion and consensus generation for these. More importantly, from a data governance perspective – to identify why a certain `code/name/test` was agreed upon and by whom. Our implementation strategy should be driven by `governance based on consensus`.
To encourage and succeed in standards adoption, we need to ensure that clinical workflows remain uninterrupted, if not enhanced. I am not referring to a workflow where a scanned PDF document from one system triggers an inbox notification in another system. Rather, a lab test result triggers a clinical decision support system where one or both these systems (at the moment) rely on `non-standard` codes/names. We might be thinking of a new LIMS system compliant with standards, but we might have no control over the other system. In fact, such data flows are common in healthcare and whatever we do we must not disrupt them. So our implementation strategy must be designed for `evolution` aka `eventual standardisation` – where different systems become standards compliant at their own pace.
Scope of Standardisation
You might have noticed that I left out my first point `new LIMS with standardisation in mind` till now. That is because I think the meaning of `standard` here needs to be defined specifically for lab medicine (pathology). I think the functional `atom/unit` in lab medicine (pathology) is a `diagnostic cycle` which combines:
- A request
- An analysis (in a lab)
- A report/result
The `why` is simple and important – the requesting clinician in many cases does not care (though sometimes does) about how a result is derived. However, the professional in the lab often cares a lot about how the result was derived – think methods, analytes, test kits, etc. The clinician again often looks for an `interpretation` by the lab specialist and corroborates it with their clinical information. However, buried in here is the notion that two tests might not always be directly `comparable` if they have been performed using different techniques, analytes, test kits, etc. Units of measure are only the most visible part of this comparability.
So unless standards address this entire functional definition, significant and differing world views between the `non-lab clinician` and the `lab professional` remain a challenge.
Addressing these would however not only address the broader operational need (remember costing is based on tests performed and the test kits used) but also make clinical data more comparable and reduce clinical risks. It also recognises the lab/pathology professionals as equal partners in the diagnostic cycle.
Enabling Implementation/Adoption of Pathology Standards
If you are with me till now, then you might be thinking – so how do we enable these aspects like `data governance`, `eventual/asynchronous standardisation` and `diagnostic cycle standardisation` which impact implementation of standards. I had a similar choice – stay on the lead of a national standard project/programme or get out and do something about enabling the adoption/implementation of these standards. So 18 months ago, I chose the latter and since then, with my colleagues at Termlex I have been working on a way to make this happen. We created:
Pathnexus – a data harmonisation platform that enables adoption of standards and addresses these issues.
- Non-standard codes (aka local codes) to be `associated` with SNOMED CT (think mapping on steroids)
- Collaborative review and consensus of both `mappings` and also SNOMED CT content that you might create in the process (topic for another time)
- Standardise both requests and results and represent the relationship between them in machine-readable formats – making operational lab analytics easier.
How it could all work?
We are now about to embark on a pilot project for testing Pathnexus, not just enabling standardisation of their lab data but also data-governance. Of course, there are the usual discussions about what is in scope for a proof of concept (PoC) –
which disciplines; how does it work with a middleware solution; is it just SNOMED CT or does it include FHIR, UCUM; what does it do with requests (standardise or not); where does governance fit?
The proof is really in making it all work. If you are curious about how Pathnexus or even the broader approach might work for you, then please get in touch. If you disagree with what I am saying or have comments about what you feel might work or needs to change, then please leave a comment or get in touch.