Anesthesia SIG (Special Interest Group) Meeting



[pic]

Anesthesia SIG (Special Interest Group) Meeting

Hotel Marienlyst, Helsingør, Denmark

Friday, 3rd April 2009

Present: Andrew Norton Anthony Madden (GoToMeeting)

Martin Hurrell Keith Naylor (GoToMeeting)

Melvin Reynolds Mitchell Berman (GoToMeeting)

Jessica Gabin Tom Marsh (GoToMeeting)

Jostein Ven Jan Wittenber (GoToMeeting)

Ronald Cornet Pippa Norton (GoToMeeting)

Alan Nicol

John Gutai

Apologies: Ron Gabel John Sum Ping

John Bastien Andrew Donovan

Jennifer Zelmer

Q1

Co-chair elections

The Anesthesia SIG has been excused co-chair elections for this cycle. The current co-chairs, Terri Monk and Andrew Norton, will remain in office until April 2010.

Minutes of the Anesthesia SIG meeting, Helsingør, October 2008

Minutes of this meeting were reviewed. It was noted that there has been a response from the American Association of Clinical Directors (AACD) regarding work on procedural times. As a result, definitions of some of the terms may be updated.

Update on preoperative dataset and SNOMED terminology (Ronald Cornet)

The group has now collected 192 data items. Each item has a Dutch translation and a category (e.g. patient history). The aim is to produce a definition for each item, a description of what it is used for, the reason for collecting it, any standard that it refers to. The group hopes to have a near-complete version by mid-April to give to Dutch anesthesiologists for feedback and further discussion. Martin Hurrell suggested that the group might like to consider as a next step the creation of use cases and HL7 CDA-compliant documents. The addition and identification of SNOMED terminology to support and populate these requirements is still at an early stage.

Q2

Joint session with nursing SIG

Catheter, cannula or line.

This was a discussion of an outstanding issue document (C00140) raised by the Anesthesia SIG and subject to joint investigation and discussion with the Nursing SIG. As previously agreed, Andrew Norton demonstrated examples of current terming and proposed revisions in the Anesthesia SIG Protégé terming application.

Two principle issues are apparent

• Where these concepts exist in procedure terms (e.g. 392230005 catheterization of vein (procedure)), there appears to be de facto acceptance of synonymy as terms such as “venous line insertion” are already present. However, not all variants may be currently represented – there is currently no “cannulation of vein”) There are some other specific vein cannulation terms for various anatomical sites (portal vein, splenic vein), where often the most common wording would be catheter – for example insertion of portal vein catheter (not a current SNOMED concept).

• In the device hierarchy, there is duplication and redundancy of concepts e.g.:

- 258620006 vascular cannula (physical object)

- 258623008 vascular catheter (physical object)

- 278422005 vascular line (physical object)

Although ISO 60601 has high level definitions that attempt to define some difference between catheter and cannula, it was agreed that at the level of practical and usable clinical terminology, such distinction is often arbitrary and not helpful. It was agreed that, for many circumstances, synonymy does exist, with a number of exceptions where a particular term is the universal term e.g. insertion of urethral catheter.

There are two possible solutions:

• Use of a word equivalence table. This would be more elegant, but would not take account of the exceptions to the general principle.

• Revision of SNOMED CT to add relevant additional synonyms in the procedural terms, and harmonise the device terms including retiral of the redundant concepts.

The latter approach needs to be scoped, as there are a considerable number of terms that will require amendment, that are not part of and will not be included in the Anesthesia SIG termset.

Action points

• Anesthesia SIG to update the issue document C00140 to reflect these conclusions and to scope the amount of work and concept modification required to close the issue.

• It was proposed that this work should be done in time for the next Content Committee conference call in 2 months’ time.

Assessment Scales

This item focussed on two areas:

• The general developing guidance for the modelling of assessment scales.

• Development of joint guidance by the Anesthesia and Nursing SIGs as to which pediatric pain scales should be recommended for inclusion and modelling in SNOMED CT.

Naming of assessment scales in SNOMED is consistently implemented. There is still debate on the modelling of components, observables and findings related to this scales.

Two new documents, posted on the Nursing SIG collaborative space, address some of the issues around this modelling. Ian Arrowsmith has drafted a tactical guideline on modelling of assessment scales for the NHS terminology centre. Andrew Norton has produced a paper on multivariate assessment scales, particularly referencing the Glasgow coma scale. (This is to be posted on the Content Committee collaborative site.) There are similar conclusions, an exception being the approach to “leaf” components of scales (the actual clinical state being observed e.g. best motor response extension to pain).

It was noted that there is work in HL7 detailed clinical models (led by William Goosen) to develop models for many of these common use scales, and these will be helpful in determining specific terminology requirements and what should be part of an information model / implementation and not the function of terminology.

As regards principles for adoption of assessment scales, the groups agreed that candidate scales should meet the criteria of widespread use, evidence based and appropriate professional authority or recommendation for their use. Anne Casey made the group aware of the Royal College of Nursing Clinical Practice Guideline “The recognition and assessment of acute pain in children” (currently in revision). This will provide a useful resource in assessing candidate scales

It was agreed that current useful candidates for adoption as pediatric pain assessment scales include visual analogue, FACES and FLACC. NIPS may also be considered.

Action Point

• Nursing SIG to consider and make recommendations on which pediatric pain scales should be considered for inclusion and modelling in SNOMED CT.

Q3/4

Joint session ISO/TC 215 WG7/CEN TC 251 WG IV/ IHTSDO Anesthesia SIG

SNOMED CT/11073 MDC co-ordination and mapping principles

Andrew Norton introduced the session, reviewing the agenda and scope for the session and what outcomes it was hoped to achieve. There were no suggested amendments to the agenda.

Melvin Reynolds gave some background to the 11073 medical device terminology work, along with some examples of the content of an HL7 (v2.6) message. Typical content includes the device and what it is measuring (each with a unique 32-bit identifier), a value for the measurement, and a time stamp. The ISO 11073 standard has the principle aims of real time records of device communication and a method for tracking and logging device error and problems

The differing purposes of existing device classifications were compared and noted to be:

Clinical : classification and recognition, low need for granularity.

Device Comms: need to communicate large amounts of highly granular information.

Regulatory: no comms requirement. Only really focused on device failure.

It was noted that the requirements of Continua for personal health device communications (that are markedly constrained by power and battery life) had led to the addition of some concepts in SNOMED CT, which although acceptable, could have been more systematically and rigorously modelled had they been subject to domain expert scrutiny.

For mapping terminology to SNOMED, it is important to agree a coordinated approach and an agreed methodology. Ad hoc mapping by different groups with different goals may lead to incorrect mapping and poor, incomplete or fragmented models within SNOMED.

Martin Hurrell briefly outlined the work IOTA has done in building a terminology for anesthesia, working with SNOMED and with 11073. He stressed the need for a coherent structure and format, rather than just a bag of terms, and pointed to the HL7 clinical document architecture as an appropriate standard for this. The HL7 Anesthesia SIG is currently working on a draft implementation guide for the anesthesia record, and this should reference the terminology needed to populate it. Provenance of terminology is important to ensure the authority of the data and to confirm that like is being compared with like.

Some 11073 terms are already in SNOMED, and where this was intentional they are referenced as such. All the 11073 terms in the IOTA Protégé tool are identified as such. He suggested that the IOTA Protégé tool could be used to import relevant areas of SNOMED terms to be tested for consistency against x73 object properties.

In any harmonisation exercise, it was agreed that the issue of attribution in both terminologies was important. John Gutai (IHTSDO Chief Technical Officer) suggested that the issue could be addressed with SNOMED RF2 (release format 2), which introduces reference sets that may allow the inclusion of cross referencing and other source information into release tables. Cross referencing information in ISO 11073 may need to include multiple SNOMED concepts, especially where equivalence is obtained by post-coordination. Jan Wittenber noted that, because of the possibly large number of SNOMED concepts and other (some manufacturer specific) equivalents, in 11073 this would be best done by constructing referenced tables similar to those used in the IHE "Rosetta" work

Action Points

• Cross-referencing between 11073 and SNOMED CT, as a mutual IP agreement, and there are issues to solve regarding attribution

• Anesthesia SIG to obtain permission to use the IOTA tool more widely than for anesthesia

• Mapping tools and process

- strongly recommend that mappings between SNOMED CT and 11073 are done using current tooling and semantic equivalence grading

- set up project proposal with IHTSDO (John Gutai to advise on this)

• Keith Naylor to liaise with Andrew Norton to identify priority areas for Continua mapping (phase 1)

The group looked at the 11073 (Annex A) standard to determine phased priority areas.

o Phase 1, use existing tooling for immediate priority maps - to be responsive to the business deadlines for publication of specifications by IHTSDO, IEEE and other organisations

o Phase 2, migrate to more recent version of Protégé for interim priority maps and then,

o Phase 3, use OWL representation of SNOMED CT for longer-term and bulk maps.

Table A.5.1 – vital signs devices. Most of these are probably too generic to be of any real use.

Table A.5.2 – some base concepts (such as analyzer, monitor) are in SNOMED, but the definitions there are not rigorous. Phase 2.

Table A.6.3 – vital signs units of measurement may be of some relevance, but the SNOMED CT use of UCUM should be investigated

Table A.7.1.2 – ECG patterns. Higher levels only are of relevance. Some new terms have been modelled in IOTA but have not yet been submitted to SNOMED, in part Phase 1

Table A.7.2.1 - ECG enumerations are a high priority and this needs to be flagged with SNOMED as an area that needs revision, phase 2.

Table A.7.3.1 – haemodynamic monitoring measurements. Second half of the table is of immediate relevance to anesthesia. Some of these terms are already in SNOMED. Phase 2

Table A.7.4 – respiratory measurements. Almost all of these have been added to SNOMED already, so the task is to add the relevant 11073 reference and 32-bit code. Phase 1

Adjourned until 4th April 2009.

Saturday 4th April 2009

Present: Andrew Norton Anthony Madden (GoToMeeting)

Melvin Reynolds Pippa Norton (GoToMeeting)

Martin Hurrell Mitchell Berman (GoToMeeting)

Jessica Gabin David Reich (GoToMeeting)

Jostein Ven Terri Monk (GoToMeeting)

Christine Spisla Jan Wittenber (GoToMeeting)

Apologies: Ron Gabel Deborah Lang-Kuitse

Steven Dain Keith Naylor

Amelia Vagnozzi John Sum Ping

Andrew Donovan

Q1

The group continued to look at the 11073 tables to sort out priority areas for mapping.

Table A.7.4.2 – ventilator modes. This is a high priority area. Some additional terms have been identified, but the whole area needs a rethink in SNOMED. The IHE Rosetta project is currently working on ventilator terminology based on 11073 and systematic analysis of ventilator function and control. There are some pre-coordinated terms already in SNOMED (e.g. controlled mandatory ventilation with sigh), but many are missing – there are none with the positive end expiratory pressure (PEEP) attribute, for example. It may, therefore, be necessary to pre-coordinate some further ventilator terms in SNOMED, but it will be best to wait for the outcome of the Rosetta project.

Table A.7.5.1 – common blood gas, blood, urine and other fluid chemistry measurements. A consistent problem in SNOMED is that there are measurement procedure terms that do not have the relevant observable entities associated with them. Sometimes the natural candidate term appears as a synonym of the procedure term (e.g. “blood arterial pH level”, a synonym of “pH measurement, arterial”). Sometimes the fully specified name itself looks more like an observable entity term (e.g. “plasma potassium level”). The IHTSDO observables project is currently tackling this issue, so generally best to wait for the outcome of this. The concern remains that this project appears to have focussed entirely on laboratory values, where as the needs for anesthesia and other healthcare devices covered by the 11073 standard have not been addressed. (Part phase 1 for Continua requirements, mainly phase 2 when IHTSDO observable remodelling test phase concluded)

Q2

Table A.7.7.2 – pump modes. Some of these terms are probably relevant. Christine Spisla asked where they would fit in the SNOMED hierarchy. Some of the terms could be findings, observables or procedures, depending on the context. Some similar terms (e.g. “triggering mode”) are already in SNOMED as “unapproved attributes”. Phase 2

Table A.7.7.3 – pump states. As very few of these would be required for the clinical record, this area is not a high priority. The terms would be more relevant for research purposes. Phase 3

Table 7.8.1 – neurological monitoring measurements. This is very detailed, but there are some crucial terms at the observation level, for example “pupillary reactions”, “circumference of head”, some of which are already in SNOMED. The higher-level evoked potential terms have been done, but not the more detailed pre-coordinated terms, such as “latency wave 4 brainstem evoked potential”. Martin Hurrell pointed out that, as there are only a few derived measures, it might be worth putting them in. It was agreed that the basic modes and the common indices should be modelled. Although a lot of the table is probably not relevant, it will be worth looking at US and UK practice to pick out relevant items. Martin Hurrell volunteered to have a first look at this. He also pointed out that the definition of “amplitude” within the table is questionable and needs clarification.

Pressure terms are required, but there was discussion as to how much detail to include. Again, as there are only a few pre-coordinated terms (mean, systolic, diastolic), it was agreed that these should be included.

The Glasgow coma score terms are of high priority, but these should be sorted out once assessment scales generally are sorted out in SNOMED. Andrew Norton has written a paper on modelling assessment scales. It is available for viewing on the Nursing SIG, Content Committee and Anesthesia SIG collaborative spaces. He would welcome any comments. Mostly Phase 3, some Phase 2

Table A.7.9.1 – neurophysiological enumerations. This is largely relevant to sleep lab work, so not relevant to the current project. It is a good reference source if the demand for it arises. Phase 3

Table A. 7.10.1 – neurophysiological stimulation modes. Again, this is not of immediate relevance, but will be useful if the demand arises. Phase 3

Table A.7.11.1 – miscellaneous measurements. Some of this (e.g. temperature site terms) is already in Protégé, though it is not yet referenced to 11073, and needs some tidying up. It was noted that patient height and weight terms in SNOMED should be considered for revision and harmonisation. Phase 2

Table A.8.2.1 – sites for neurophysiological signal monitoring: locations near peripheral nerves. These are referenced from those approved by the International Congress of Anatomists. It was noted that virtually all the terms already have a SNOMED code, albeit the old RT codes. Christine Spisla will look into a way of pulling these terms out of the current SNOMED to do a sample check of cross-mapping. Phase 2/3?

Table A.8.9.1 – qualifiers of body site locations. It will be necessary to talk to anatomy experts (possibly Kent Spackman) for guidance on how to manage this area. The 11073 terminology pre-coordinates left, right, distal, proximal. Phase 2/3?

Table A.9.2.1 – pattern events. Some of these have been modelled into Protégé, and some are already in SNOMED. ECG pattern findings have been revised by IOTA and some new terms have been identified. This is all pretty well worked out and cross-mapped in Protégé. It is an area that needs a lot of work in SNOMED. Andrew Norton will submit the IOTA model. Phase 2

Table A.9.3.1 – device-related and environment-related events. There are some general terms, such as “equipment error” in SNOMED, but most of the 11073 terms will not be needed. Some extra terms have been identified for anesthesia incident reporting. Phase 2/3?

Melvin Reynolds suggested that it may be possible to get some funding for this mapping project through the European Commission. There is an existing proposal from 11073 regarding maintenance of 11073 terms, and it may well be possible to revamp this proposal to include the Protégé tool.

Q3

Report from the IHTSDO Content Committee, 2nd April (Andrew Norton)

There is an IHTSDO terminology modelling workbench in development but from experience at a tutorial earlier in the week does not yet appear easily usable for terminology developers who were not receiving significant support from IHTSDO.

The test phase for the observables project will focus on laboratory procedures. While this is an important area, it is felt that the focus really needs to be wider.

Stefan Schulz gave a presentation comparing the ontology basis of SNOMED CT with the granularity of the vocabulary, and the problems that might potentially arise from this. It may be particularly relevant with regard to US regulatory terms where there is potential for combinatorial explosion. Andrew will try to obtain a copy of the presentation to post on the Anesthesia SIG collaborative space.

Substance hierarchy model for non-depolarising muscle relaxants

There are inconsistencies in this area in SNOMED and some reorganization is needed.

Andrew Norton showed the proposed Protégé model to the group. “Neuromuscular blocking agent” has two children, “depolarizing agent” and “nondepolarizing agent.” Nondepolarizing agents are split into three groups: “benzylisoquinolinium” (replacing the erroneous “benzoquinonium” in SNOMED, “curare and derivatives” and “steroidal neuromuscular blockade”. It was agreed that the Protégé model looks complete and correct, and that it should be submitted to SNOMED.

Substance hierarchy model for cyclodextrins (sugammadex)

Sugammadex is already licensed in the UK and other European countries, it therefore needs to be modelled into SNOMED, but it is not clear where it fits in the existing hierarchy. Apparently it has gone into the UK extension, but nobody in this group has seen the latest UK extension, so it is unclear how this has been done or where it appears. Mitch Berman reported that FDA approval is imminent.

There are three classes of cyclodextrins: alpha, beta and gamma cyclodextrins. They are classed as antidotes according to the World Health Organization classification. The term “antidote” exists in SNOMED, and it was agreed that this would be fine as a functional classification for sugammadex. Ideally, it should have a chemical classification. A proposed model for this was created in Protégé, using the existing SNOMED “carbohydrate” (modelled via “organic compound” rather than “nutrient”):

organic compound

carbohydrate

polysaccharide

cyclodextrin

alpha cyclodextrin

beta cyclodextrin

gamma cyclodextrin

sugammadex

Q4

Guidance for modelling of assessment scales

The group looked at a guidance document from the UK Terminology Centre, which will be posted on the Anesthesia SIG collaborative space. The document is not easy to follow and understand, particularly as there are no examples to illustrate the points made. However, there are some areas of concern. For example, the paper states that observations related to leaf-level component scores will not be routinely modelled in SNOMED. This group feels that these observations should be routinely modelled unless it would not be possible to develop meaningful SNOMED concepts. This has been done in Protégé and SNOMED CT for Mallampati and Cormack and Lehane scores.

The group then looked briefly at Andrew Norton’s discussion paper on modelling assessment scales. The Glasgow coma score is given as an example of a multivariate scale. SNOMED has pre-coordinated findings terms for all the scores (e.g. Glasgow coma scale, 3 (finding)). However, this doesn’t indicate what the different component scores are that make up the total score. There are subscore terms held as qualifier values (e.g. Glasgow Coma Score (GCS) eye opening subscore (qualifier value)). The leaf concepts for each could be held as clinical findings (e.g. eyes opening spontaneously, to speech, etc.).

Cormack and Lehane is used as an example of a univariate scale. SNOMED currently holds pre-coordinated terms for the grades (e.g. cormack and Lehane grade 1 (finding)). The pre-coordination of values is discouraged. An alternative method might be to interpret the score from the leaf concepts.

This document will be available on the Anesthesia SIG collaborative space, and discussion and comment is invited.

Martin Hurrell suggested that a worthwhile exercise to run in parallel with this would be to assign appropriate properties to the terms in the current termset. A number of classifiers could then be used to throw up different possible hierarchies based on the properties. This would give a flexible tool for looking at the ontology in parallel with the development of terms. Martin would be happy to look at putting the tools together for this and making a start on it.

Update on procedural times

Procedure milestones have now been modelled into Protégé, along with definitions, and are in the submissions database. They are based on terms from the American Association of Clinical Directors (AACD), Royal College of Anaesthetists and NHS Scotland Clinical Dataset Development Project. Some reworking has been done to harmonise with SNOMED Ct conventions and to reflect multiprofessionalism in anesthesia. It has been brought to the Anesthesia SIG’s attention, via the Association of Operating Room Nurses (AORN), that AACD are going to relook some of their terms. It is understood that basic times will be left alone, but some of the supporting definitions may change. It may therefore be best for the Anesthesia SIG to hold back on the submission of their terms until proposed changes are known.

Observables for blood gas analysis

“blood gas analysis (procedure)” exists in SNOMED, but relevant observable entity terms are missing. There is “blood glucose concentration” under “sample observable”, but it isn’t really in harmony with the rest of the terms there. Andrew Norton will send Christine Spisla a list of problematic observables terms to consider where they might appropriately fit into the SNOMED hierarchy.

Analysis of device output nomenclature

Jan Wittenber talked the group through work he has done on trying to match some 11073 terms with IHE PCD Rosetta terminology, and the problems that arise from variations due to how manufacturers define their terms. A large number of terms can be reduced to a much smaller set once the variations are filtered out and equivalent terms are recognized as such. To be certain that terms really are equivalent, however, it is necessary to have input from someone who understands the devices aspect and what they’re measuring, and a medical expert in the relevant field. Some gas monitoring terms were given as examples: expired CO2 and end tidal CO2 – are these really different? If so, are they significantly different clinically, and in what context? It is necessary to have a clear definition and explanation of what the differences are where they are important, in order to disambiguate the terms in a clinically meaningful way.

A harmonized set is needed, that is normative, encompassing 11073, LOINC, etc. There’s also a need for complimentarity. If some definitions are not appropriate within the SNOMED model, 11073 can work on them and vice versa. Modal and measurement site attributes are difficult to deal with and both 11073 and SNOMED need to do some work on these areas. There should be a similar set of attributes, as harmony at this level would make higher-level, composite terms more similar.

Melvin Reynolds outlined the three phases of mapping work proposed by this group to which Jan’s work is highly relevant. Phase 1 is areas considered urgent, which will be done on a precipitate one-off basis. Phase 2 is areas where demand will exist in the near future. Phase 3 is areas which will be addressed if demand arises. It is hoped that development of the Protégé tool will make the mapping of terms into SNOMED a more automatic process, which will facilitate work in phase 2 and 3.

Action Points

• Martin Hurrell to look at the 11073 neurological monitoring measurements terminology and pull out relevant modes and indexes for modelling into Protégé and SNOMED.

• Christine Spisla will look at using the old SNOMED RT codes to pull out existing neurophysiological monitoring sites from the current SNOMED CT.

• Andrew Norton to submit to SNOMED the Protégé model for ECG pattern findings.

• Melvin Reynolds to investigate the possibility of some funding from the European Commission for this mapping project.

• Andrew Norton to post a copy of Stefan Schulz’s presentation to the Content Committee on the Anesthesia SIG collaborative space.

• Andrew Norton to submit to SNOMED the Protégé model for non-depolarising muscle relaxants, and the suggested model for cyclodextrins and sugammadex.

• Christine Spisla to provide access to the Anesthesia SIG collaborative space for Melvin Reynolds, Jessica Gabin and Jostein Ven.

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download