College bows out of SNOMED development

1

Do we know the risks?

11 Jan 07 08:36

In my opinion, this is a worrying development that could leave the NHS seriously exposed. The announcement - in 1999 - of a 20 year contract between the NHS and the College of American Pathologists implied commitment, stability and enough confidence to embed SNOMED (instead of Read Codes) in NHS systems. On the other hand, the Standards Development Organisation proposal appears speculative in comparison. Last time I asked CfH, they had no business case for this initiative: if one has now been drafted then it would be interesting to see it.

There are four other points that arise from the article that need to be clarified:

1. SNOMED may well be "used" in 30 countries - but why, then, did only 5 meet last autumn to discuss the formation of the SDO? (http://www.e-health-insider.com/news/item.cfm?ID=2250). And just 4 others have subsequently expressed an interest.

2. The NLM agreement allows for SNOMED to be available in the USA as part of UMLS, along with many other terminologies and classifications. The NHS - on the other hand - appears to have all its terminological eggs in just the one basket.

3. International standards initiatives in health IT have historically been characteristed by inertia, bureaucratic consensus processes, and "slowness-to-market" of their initiatives. Is this an ideal model for supporting the NHS's dynamic terminology requirements?

4. Richard Granger's aspirations for global data need to be clarified. In a survey of around 150 international terminology experts I undertook in 2001, there was recognition that comparison of data internationally was important, though transferring clinical data to support continuity of care was not rated as significant. The fact that we've not yet succeeded (on a useful scale) in transferring electronic patient data across market towns in the UK reinforces this finding and - perhaps - suggests that both Mr Granger and the UK terminology community should focus closer to home.

Terminology initiatives over the past 15 years (Read v3, GALEN, SNOMED RT and SNOMED CT) have produced some impressive scientific and technical advances. The real challenge, though, has been creating the business model that supports and maintains the results of these efforts. From the evidence of which I'm aware, the combination of: technical complexity + global consensus processes + linguistic differences + different healthcare delivery models + financing the process, militates against the proposed approach.

Colin Price

---

2

Professor Rector's assertion

rf@medicineit.com

11 Jan 07 10:27

I like the concept discussed by Professor Rector at last Year's Implementing SNOMED CT conference in London:

SNOMED CT + HL7 = CHAOS

There is no data to demonstrate the implementability of SNOMED CT on a scale as large as the NHS is attempting. As Professor Rector also stated at the same conference (I paraphrase): "Just because people tell you that it will work doesn't mean that it actually will work".

We have no evidence that this will work in a busy user environment and there are people jetting around the world excited by all of this! It would be nice to see these people actually delivering this into the user environment where it is supposed to be used rather than pontificating academic superiority over the rest of us. Then they can jet off wherever they wish to go and be proud of their actual achievements.

Academic definition of a language is the first step in making it a sustainable communication medium. There are plenty of opportunities for this to fail yet!

Time is a resource that often purifies the truth from fiction!

---

3

Misleading article title

11 Jan 07 17:25

The understanding given at all recent SNOMED meetings is that the College of American Pathologists would continue to author and maintain the SNOMED codes for at least a couple of years after the inception of the Stategic Development Organization. (After that period the maintanance contract would go out to tender).

CAP are proposing to relinquish ownership of the intellectual property to the SDO along with the expense of maintaining it! They will still be *developing* it. This however raises complex questions over paying of pipers and calling of tunes.

It is *very* labour intensive to maintain a would-be totally comprehensive clinical terminology. CAP are merely doing in their turn what the NHS tried to do to them with Read/Clinical Terms - offload the cost onto another organisation.

Meanwhile SNOMED continues with unresolved QC and (much more worryingly to early adopters) knowledge representation issues.

It is therefore a shame that politicing over the SDO has led to the cancellation of the February SNOMED editorial board and working group sessions.

Some EHI readers may think these issues peripheral to NPfIT - think again! In particular NCRS is wholly predicated on the universal use of SNOMED across the NHS.

The near total lack of SNOMED enabled systems on offer to the NHS - four years after an (admittedly toothless) mandate, and three years into NPfIT should raise interesting questions at the highest levels. The OBS stipulated such systems should be ready by Dec 2004!

---

4

the challenge facing SNOMED goes beyond the SDO

ben.toth@gmail.com

15 Jan 07 05:52

I agree that the transfer of ownership to the SDO presents a risk to its future viability. But there is a greater risk I believe, which is that there is no convincing practical demonstration of (or even busines case for) the benefits of SNOMED CT. This is needed to sustain what will be a challenging implementation process.

---

5

Implementability

17 Jan 07 08:14

Ben Toth makes a valid point (echoing an earlier comment in this thread) about the implementability of SNOMED CT. The notion that a large comprehensive terminology was needed in the NHS came about in the early 1990s, giving rise to the Clinical Terms Project (CTP), and subsequently becoming perpetuated as part of informatics "folklore". Intuitively, it does seem like a sound idea and therein lies a danger: in the belief that both the benefits and technical feasibility are self-evident, the DH has gained support for an NCRS founded on a large shared vocabulary.

A 1999 Business Case was produced for the further *development* of clinical terms for the NHS; but a range of issues continue to raise concerns about actual implementation:

1. The CTP began in April 1992. Fifteen years later, the NHS front line has seen little or no practical impact from this or its successors: certainly not enough to justify the costs incurred to date.

2. Coding schemes have achieved some useful penetration in UK primary care (Read Codes) and in pathology (SNOMED), perhaps due the nature of their records. Evidence of large-scale (particulaly national level or interdisciplinary) adoption in other sectors is hard to find.

3. The Public Accounts Committee recognised almost 10 years ago the risks inherent in this "terminology proposition" leading to reassurances that there would be rigorous evaluation before NHS implementation and, indeed, the DH went on to commission two evaluation methodologies (technical and economic) from an international team of experts. At that time, the PAC's focus was on Read Codes version 3: it's hard to see how the switch to SNOMED, of itself, obviated this type of evaluation.

4. We need to understand the added costs that result from internationalisation per se of terminology development. In evidence to the US National Committee on Vital and Health Statistics in 2002, I estimated that such an endeavour might cost around $15 million per annum. At that the time, the UK's own terminology service within the NHSIA was costing only in the region of £400 - £500k per annum.

In my opinion, there are two key questions that remain unanswered:

(a) Given that "seeing is believing", do we have proof of concept for the large-scale multi-disciplinary terminology proposition?

(b) If so - and secondarily - does SNOMED CT supported by a SDO meet the technical, useability, value-for-money and long-term business viability criteria to form the basis of NCRS as currently envisaged?

My view on both remains neutral, but - given the proposed change to a SDO and what appears to be a significant scaling-down of NCRS content - now seems like an ideal time to attempt to try and find some answers.

Colin Price