SNOMED CT. should Sweden join now or wait?

Transcription

1 SNOMED CT should Sweden join now or wait?

2 The National Board of Health and Welfare categorises its publications according to document type. This is a Status Report. This means that it contains a report and an analysis of surveys and other means of following up legislation, activities, resources and so on which municipal authorities, county councils and individual principals utilise and pursue within the health service, social services, health protection and infectious disease control. It could constitute a basis for decisions reached by the Board and could form part of a more extensive follow-up and evaluation of, for example, reforms and allocation of stimulation funds. The National Board of Health and Welfare is responsible for the contents and conclusions. Artikelnr Publicerad november

3 Foreword Almost 40 years have passed since the first computer systems came into use in the health service. Both then and now it is the art of describing the content of the health service system that accounts for a great deal of the development work. The focus of development work during the 21st century has been on patient safety, compatible data and communication between computers. These are areas in which it is assumed that there are national definitions and agreements regarding concepts, terminology and information structure, i.e. the core of all computer systems. Although there are several excellent dictionaries with medical terminology there has not been any equivalent for use in computer systems. Globalisation also means that concepts, terms and information structure should be harmonised with other languages. A British-American concept system, SNOMED CT, is the most complete system available today. Should Sweden be involved in the future development of this concept system? This status report contains a summary of the documentation to be found in SNOMED CT, interviews with some of the organisations and persons who have exerted pressure to induce the National Board of Health and Welfare to accept responsibility in this matter as well as the considerations of the working group that was appointed. The assignment co-ordinator was Lotta Holm Sjögren, project manager, Carelink. The material for the report was produced by Anna Vikström, doctoral student, Karolinska Institute, Daniel Karlsson, PhD, research assistant, Linköping University and Östergötland County Council, Staffan Bryngelsson, consultant, Emendor Consulting AB, Hans Åhlfeldt, Professor, Linköping University, Lars Berg, Medical Director, National Board of Health and Welfare, Ulla Gerdin, terminology co-ordinator, National Board of Health and Welfare and Boel Engarås, project manager, National Board of Health and Welfare (up to ). The report was compiled by Lotta Holm Sjögren, Lars Berg, Stefano Testi, terminologist, National Board of Health and Welfare and Ulla Gerdin. Petra Otterblad Olausson Head of Department Centre for Epidemiology 3

4 4

5 Contents Foreword...3 Summary...7 Introduction...9 The assignment and its purpose...9 Background to the assignment...9 Content and the limitations of the assignment...11 Method...12 Explanations of words...12 Need for uniform information...13 Documentation in health care...13 Terminology work at the National Board of Health and Welfare...13 Standardised use of concepts and terms...14 Regional and local terminology work...14 Classifications within the health service...14 Uniform national information structure...15 Project on the way to becoming a uniform information structure...15 Expectations of national terminology for the health service...18 Interviews...18 Results of the interviews...19 What is SNOMED CT?...20 History...20 The purpose of SNOMED CT...20 Facts about SNOMED CT...21 Possible areas of application...24 SNOMED CT in comparison...26 SNOMED CT, the National Board of Health and Welfare's termbank and WHO classifications...26 Information models and concept systems...29 Evaluations and applications...30 Scientific studies of the content of SNOMED CT...30 Applications...33 SNOMED CT ownership and management

6 The organisation behind SNOMED CT...36 Offer of participation in SSDO...36 Facts about SSDO...36 Formation of SSDO, status June Known costs...41 Considerations...42 References

7 Summary One of the necessary components for good, safe care is that the information that is communicated between people (patients, users and health service personnel), to computers and between computers is uniform and clear and that the context and content are not lost in the communication process. The National Board of Health and Welfare set up a working group in April 2006 to examine the question of Swedish involvement in the future development of the concept system SNOMED CT (Systemized Nomenclature of Medicine, Clinical Terms). The inquiry came about following an invitation sent by the College of American Pathologists (CAP) and National Health Service, Connecting for Health (NHS CfH) in the UK, to a large number of countries regarding a new international organisation for the further development and management of SNOMED CT. The organisation is known as SNOMED Standards Development Organisation (SSDO). The inquiry is based primarily on material that is in the public domain and which has been published scientific articles, conference reports, evaluations and project reports as well as material from WHO (World Health Organization), educational visits to Denmark and material from the SNOMED CT organisation. The inquiry group also interviewed representatives from the health service and universities. The inquiry describes the need for a common concept system in the Swedish health service and expectations about what such a system should contribute. Also reported are facts about SNOMED CT and international experience in the form of applications and evaluations. Sweden as a nation must decide which path it will follow, where the alternatives are either to become involved in collaboration surrounding SNOMED CT or follow its own line. The choice of direction must be linked to the desired level of ambition in the future use of information systems within the health service and what time frame we have with regard to development. Among our neighbouring Nordic countries, Denmark is part of the international SNOMED CT collaboration. Finland, Iceland and Norway have yet to decide. There is a great lack of published usable references related to the practical use of SNOMED CT that could act as guidance in conjunction with implementation. SNOMED CT is the concept system which currently has the broadest coverage but despite this it cannot satisfy all needs within the health service. Becoming part of the international SNOMED CT collaboration should be seen as a development project in itself, which is in turn part of other national development work. If SNOMED CT is to be used, it must be translated into Swedish and adapted to Swedish conditions. 7

8 National translation and adaptation of SNOMED CT must be done in such a way that it also satisfies the local, national and Nordic terminology and classification requirements. SNOMED CT is not a replacement for future work on classifications and linguistic definitions of concepts and terms. Without the aid of computers and modelling of relations, the design and maintenance of SNOMED CT is a practical impossibility. Working with SNOMED CT requires knowledge and resources for management and maintenance on the national, regional and local level. A low level of knowledge of SNOMED CT together with unrealistic expectations or an underestimation of the work required on different levels are aspects that must be taken into account in the planning if SNOMED CT is to be introduced and used in the Swedish health service. The inquiry group has made an assessment that a concept system equivalent to a "complete" SNOMED CT will be needed if the health service is to develop and if the use of information systems is to be effective and positive in the long term. The people who were interviewed support this assessment in their statements. There is also a broad consensus that there is a need for national responsibility and a uniform concept apparatus for health service documentation. 8

9 Introduction The assignment and its purpose In April 2006, the National Board of Health and Welfare appointed a working group to examine the question of Swedish involvement in international collaboration surrounding SNOMED CT (Systemized Nomenclature of Medicine, Clinical Terms). Such involvement would entail participation in the future development, use and management of SNOMED CT and the concrete task of adapting and introducing the system into the Swedish health service. The purpose of the assignment has been to provide the National Board of Health and Welfare with a basis for making an assessment of whether Sweden should be involved in the development, ownership and management of SNOMED CT. Although it was planned that the enquiry would be completed in June, the time was extended to August Background to the assignment In June 2005 a project group from the College of American Pathologists (CAP) in the USA and National Health Service, Connecting for Health (NHS CfH) in the UK formulated a proposal for a new organisation for the development and management of SNOMED CT. The new organisation is called SNOMED Standards Development Organisation (SSDO). Since November 1, 2005, a large number of countries have been invited to become members of SSDO. In recent years several national initiatives have been taken in Sweden with regard to IT development in the health service. A number of these initiatives have strong links to terminology work and the development of a common information structure. Below are some examples. National IT strategy for the health service On the initiative of the Ministry of Health and Social Affairs, a national management group for IT in the health service 1 was set up in March In March 2006 the management group presented a national IT strategy for the health service [1]. The strategy presents six focal areas where national co-ordination is required to improve efficiency in the use of IT in the health service and create conditions for a safe, available, efficient health service system. One of the focal areas describes the need for a common information 1 The national management group for IT in the health service is made up of representatives from the Ministry of Health and Social Affairs, the Swedish Association of Local Authorities and Regions, the National Board of Health and Welfare, the Medical Products Agency, Apoteket AB and Carelink. 9

10 structure, i.e. a national set of rules which describe the content, format and structure of the information so that it can be handled in the health service IT system. Government assignment regarding the standardised use of concepts and terms In 2005, the government charged the National Board of Health and Welfare with the task of "standardising the use of national terms and concepts and producing a uniform information structure within the health service aimed at creating clear information that supports communication and interaction between principals". The assignment has its starting point in the knowledge and experience generated through the project Informationsförsörjning och verksamhetsuppföljning i vård och omsorg (InfoVU) (Provision of Information and Operational Follow-up in the Health Service) run between 2001 and 2004 [2]. In April 2006, the National Board of Health and Welfare presented the report Normerad användning av begrepp och termer och enhetlig informationsstruktur inom vård och omsorg (Standardised Use of Concepts and Terms and a Uniform Information Structure within The Health Service) [3] to the government. The report highlights the conditions which the National Board of Health and Welfare needs to standardise within the area of concepts, terms and information structure. There is also a presentation of the considerations and proposals which the National Board of Health and Welfare felt were important pending a decision regarding standardisation. The National Board of Health and Welfare has also produced Handbok för arbete med begrepp och termer (Handbook for Work on Concepts and Terms) [4]. The handbook is an aid in the National Board of Health and Welfare's internal terminology work. The aim is also to work to induce parties involved in national terminology work in the health service to use common methods and working forms. In the above-mentioned report [3] the National Board of Health and Welfare also presents proposals for the build-up and operation of a National Information Structure Library. The report describes the conditions for how the information specifications produced would be made available and managed after they have been accepted, examined and quality-assured. The proposed national library will be analysed during autumn 2006 based on the scope, responsibility, costs, effects and possible risks associated with the introduction of the library. The results of the analysis will be used as a basis for a decision about the library. Government assignment, national information structure 2006 In August 2006, the National Board of Health and Welfare presented the government with Handlingsplan för Nationell informationsstruktur för individbaserad ändamålsenlig vård- och omsorgsdokumentation (Action Plan for a National Information Structure for Individual-based, Dedicated Health Service Documentation) [50]. The National Board of Health and Welfare assumes strategic responsibility for the co-ordination of a national informa- 10

11 tion structure. The assignment will be implemented in close collaboration with the principal and other key players. The national IT strategy for the health service is the basis of the vision and the common objectives that will form the starting point for this work. The work involved in setting up the national information structure will also in the best way possible build on work done previously within, for example, InfoVU [2]. Proposal from principals and universities In a letter [5] to the Director-General of the National Board of Health and Welfare, dated October 2005, the health service principles and universities in the Skåne region, Stockholm County Council and the Västra Götaland Region emphasised the importance of the National Board of Health and Welfare leading and hastening the strategic work on the creation of a uniform terminology and information structure for the Swedish health service. It is also recommended that the National Board of Health and Welfare examines as soon as possible the question of the possible introduction of SNOMED CT. Content and the limitations of the assignment The assignment has involved compiling available documentation information about SNOMED CT and SSDO. The results of a number of interviews have also been compiled. The assignment also included considering and recommending a position from Sweden's point of view regarding participation in the future development of SNOMED CT. The report highlights the following areas: The need for a common concept system in the Swedish health service. Expectations of SNOMED CT Facts about SNOMED CT function, use, ownership and management. Comparison SNOMED CT in relation to the WHO's classifications and the National Board of Health and Welfare's termbank. Experience of SNOMED CT use and evaluation. Considerations join the SNOMED CT work now or wait? A decision about using SNOMED CT in the Swedish health service entails costs in various forms. The inquiry has in the report specified known national costs for a licence and so on. Other costs for the development, adaptation and introduction of SNOMED CT in health data registers, regional and local health service registers, health record systems etc. have not been calculated. Nor have costs for national co-ordination, build-up of know-how, organisation and management been calculated as this requires work which has not been included in the framework of the assignment. Certain national costs for translation etc. have been given, based on experience in Denmark. 11

12 Method Documented information The report is based mainly on the collection of various kinds of material already published and in the public domain scientific articles, conference reports, evaluations and project reports as well as material from the SNOMED CT organisation. Material has also been gathered from databases such as PubMed and Cinahl as well as through searches in Google Scholar and Google. Personal contact has provided access to material from, among others, WHO in Denmark. The material used in the inquiry is reported in the reference list. Interviews Interviews with representatives from the health service and universities were interviewed by the inquiry group. The interviewees were in the first instance appointed following proposals from those persons who in October 2005 signed the letters to the Director-General of the National Board of Health and Welfare regarding the importance of hastening the work related to common terminology and information structure for the Swedish health service. A total of nine persons were interviewed. Educational visit to Denmark In Denmark, a national and regional health records project is in progress where SNOMED CT is used as an aid in achieving the aims and objectives of the project. As the project, which has been running for a couple of years, and the Danish National Board of Health are considered to have experience of significance to this inquiry, the inquiry group arranged an educational visit to Denmark where the H:S Hovedstaden region and the National Health Board presented their work. Explanations of words In this document the concepts below are used with the following meaning: terminology: set of terms within a subject field (Source: National Board of Health and Welfare termbank). concept system: set of concepts associated with concept relations (Source: National Board of Health and Welfare termbank). classification: the result of objects or individuals being grouped, divided or arranged into different classes (Source: National Board of Health and Welfare termbank). 12

13 Need for uniform information Documentation in health care In the health service, information is required in order to make decisions regarding individual patients or users and to facilitate administration in conjunction with the health service, control, follow-up, development and research. The information must be a long-term resource, independent of any organisation or technical solutions. Several national initiatives have been taken in recent years to satisfy the need for uniform, unambiguous information. Below are some of the objectives that have been formulated and which are linked to information and the use of information: "Reliable, usable health service information easily accessible to the general public, employees and decision-makers" [2]. "There are definitions of the concepts used in operational follow-up. The definitions are available in the termbank" [2]. "Provide the general public and patients with comprehensive, readily available information about access to the health service and the quality and results of the health service" [1]. "Create the opportunity to access information about one's own care and health situation" [1]. Information is communicated with the aid of concepts and terms found in a specific context. The vision is that the same concepts should be interpreted in a uniform, unambiguous way regardless of whether they are communicated by computers or by people. The content of the concept and its context should not be lost in the transfer. Terminology work at the National Board of Health and Welfare The aim of the terminology work at the National Board of Health and Welfare is to create a common special language for health service. To achieve this, the concept definitions must, as far as possible, be general, i.e. they must be interdisciplinarily applicable to the professional groups and operating areas concerned. The focus in the terminology work is national co-ordination within the health service. The National Board of Health and Welfare organises and 13

14 carries on terminology work based on established methods and a clear decision-making process in collaboration with different parties. The National Board of Health and Welfare is working to increase awareness of the significance of uniform concepts and terms among health service personnel. Concepts and terms used by the National Board of Health and Welfare in, for example, instructions, guidelines and on the website, as well as in communication with the general public, other authorities and organisations, should be uniform and unambiguous. The National Board of Health and Welfare provides and manages a public, web-based termbank that contains recommended concepts and terms. Standardised use of concepts and terms The National Board of Health and Welfare has the potential to standardise and recommend the use of concepts and terms which are of significance in protecting the individual, for operational follow-up and planning and for the collection of health service data, quality data etc. [3]. The aim of the standardised use of concepts and terms and a uniform information structure is that the parties involved in the health service co-operate, describe and use concepts, terms and information structures in a uniform, unambiguous way to achieve safety, efficiency and quality in the handling of information. Examples of areas in which standardisation could be relevant are the use of concepts and terms that are of significance to the safety of the individual, both in legal terms and as a patient, care of the elderly, the design and content of health records, quality management, the use of classifications and other codification as well as health data registers. Standardising or recommending the use [3] of concepts and terms are processes that require terminology work and acceptance and take time. Regional and local terminology work To achieve these objectives, concepts and terms must be used uniformly on all levels in the health service. In conjunction with the local and regional introduction of IT systems within the health service, concepts and terms are being developed to satisfy the purpose of the local project [2]. In connection with this, national sources are used such as the National Board of Health and Welfare termbank and existing classifications. The termbank contains medical concepts to a limited extent but still lacks many of the concepts and terms required for good health service documentation. The demand for nationally adopted terminologies is clearly greater than the supply. Classifications within the health service Classifications are required within the health service in order to be able to follow up statistically reasons for contact with the health service system (reason for encounter) diseases and health problems (diagnoses) [6] consequences of these diseases and health problems (functioning) [7] interventions (procedures) [8] 14

15 The use of established, common classification systems creates the conditions for comparing results or relating results to preset objectives. Primary classification means, for example, that the diseases and health problems at a contact with health care is coded with the aid of a diagnosis classification. Secondary patient classification means that an encounter is placed in a certain category with the aid of data from one or several primary patient classifications. DRG is an example of a system for secondary patient classification, where each time care is provided it is sorted into a certain group, depending mainly on primary classified diagnoses and procedures. Uniform national information structure A uniform information structure means that the information used in computer systems in the health service is uniform and is described in full, has a clear structure and can be reused. The information structure contains necessary concepts and terms, classifications and other code systems as well as rules for the information that is to be processed. It also contains a machinereadable description of the contents of the information model. The need for a uniform national information structure is emphasised in the national IT strategy for the health service: "A uniform, national information structure based on standardised use of terms, concepts and classifications is an important basic pre-requisite for both patient safety and the follow-up of the health service" [1]. The national requirements regarding data quality in IT systems have not previously been formulated for the health service. Concepts and terms, classifications and IT systems have been developed individually. There has been a lack of a common, uniform national information structure based on operational information requirements. The information structure comprises several interacting elements. Concepts and terms make up one of these elements. The work on different parts of a uniform information structure must take place through co-operation between many different players the county council, municipal authorities, private healthcare providers, authorities, universities and other organisations within the health service (specialist associations, professional associations etc.). Project on the way to becoming a uniform information structure Several current national projects aim to make health service information available and make it into a joint resource through the production of information specifications structuring and describing information that is documented in different contexts. This project work is an important part of the gradual development of a national information structure for the health service. Below are examples of national projects that are contributing to a national information structure. For project work there is method support, RIV (Rules for electronic interoperability in the health service) [9], a set of rules 15

16 which have been developed by Carelink in co-operation with the National Board of Health and Welfare. Project Information Structure for Quality Registers The Information Structure for Quality Registers project (IFK) [10] is a pilot project aimed at commencing adaptation of quality registers to a uniform national information structure in order to facilitate data transfer and reduce duplication of input. The project is a collaborative venture between the Swedish Association of Local Authorities and Regions, the National Board of Health and Welfare and the Västra Götaland Region. Information currently found in national quality registers is not presented uniformly and does not have a common structure with regard to concepts, relationships and rules. The technical solutions and associated security solutions vary. This means that those areas that wish to develop integration between registers and health records must use different technical solutions for each register and health record system. The overall purpose of the project is to ensure that the provision of information for national quality registers and health record systems can take place without registration being duplicated different health record systems and different national quality registers only need one adaptation for mutual interaction. In addition to the information structure, a common base function specification is required to ensure that the exchange of information between health record systems and quality registers takes place in an effective, qualityassured manner. Project National Patient Overview The National Patient Overview project (NPÖ) [11] was run in 2005 at the request of the county council directors. Carelink was responsible for the project management and the aim was to produce and test a pilot version of a national patient overview. Security in the pilot version was based on existing authorisation control systems with added security between the different services (e.g. consent service, logging service). The project was based on the assumption that the health service information needs to be divided between players from different care providers in a care chain and also be made available to the patient to increase the patient's involvement in his/her own healthcare. When choosing the content of the pilot version the starting point was the information that was regarded as being of greatest benefit to the users, or the information that was available in IT support at the four pilot county councils and the information that was possible to provide with a sufficient level of quality. Examples of information in the pilot solution are the patient's basic data, primary care contact in the patient's home area, laboratory results, prescribed medicines and diagnoses. The project also made an inventory of information areas that had been gone through thoroughly using sustainable terminology and information 16

17 models. The results were thin. The solution for the pilot version was to produce information in the form found in the health service systems and package it into a format based on available standards. Harmonising underlying structures was not possible at the pilot development stage. A key task in future work on the National Patient Overview is thus to pursue the development of a uniform information structure. Child Health Data project In spring 2006, the Swedish Association of Local Authorities and Regions initiated a Child Health Data project [12] aimed at developing, co-ordinating and improving the efficiency of information processing in paediatric healthcare and the school health service. The aim is a nationally agreed model for the electronic maintenance of health records and the transfer of information between the two areas. The project will have completed the bulk of its assignment during The national model will include a nationally agreed demand specification, which municipal authorities and county councils can use in the procurement of future medical record systems for paediatric healthcare and the school health service. The following should be presented: The nature of the health-oriented information about each child that should be registered in the paediatric healthcare and school health service systems. How this information should be registered and how it can be transferred within and between the principals' operations. How the IT support required within the paediatric healthcare and school health service systems will function. The information that is registered about each child should follow/be entered into the national information structure for which the National Board of Health and Welfare has strategic responsibility [13] and is based on terms and concepts that have been defined in the National Board of Health and Welfare termbank. 17

18 Expectations of national terminology for the health service Interviews Part of the work of investigating the question of Swedish involvement in international co-operation related to the SNOMED CT terminology has been to interview the persons representing health service principals and universities in Stockholm, Västra Götaland and Skåne who signed the letter sent to the National Board of Health and Welfare in October 2005 [5]. Those who signed the letter personally selected the interviewees. Göteborg University was not represented in the interviews due to the fact that the working group did not receive the name of the person to contact in time. Representatives from two other county councils, Värmland and Halland, were also interviewed. These persons were chosen by the working group. Another county council representative (for Norrbotten) was contacted but without it being possible to conduct an interview. An interview was also conducted with the head of the Nordic Classification Centre in Uppsala. The persons interviewed were: Martti Virtanen Centre Head Nordic Classification Centre in Uppsala Barbro Naroskyin Director GVD (Joint Health Service Documentation) programme, Stockholm County Council Ulf Jacobsson IT Director Stockholm County Council Göran Elinder Professor Karolinska Institute Peter Lönnroth Deputy Health Service Director Västra Götaland Region Jan Nilsson Professor, former Dean Lund University Jarl Lichtenberg Head of Department Skåne Region Lisbeth Lindahl Project Manager Halland County Council Lars Midbøe Operational Developer Värmland County Council The interviews were open. The opening question put to all interviewees was: "What expectations and needs exist from your organisation's point of view for national health service terminology?" Other questions in the interview had their starting point in how each interviewee replied to the opening question. 18

19 Results of the interviews All interviewees stated that it is important/absolutely necessary that responsibility for the work involving "uniform clinical terminology" should be on the national level. The majority also stated clearly that this responsibility rests with the National Board of Health and Welfare. "If the National Board of Health and Welfare does not take or is not given the opportunity to take this responsibility, those who signed the letter will probably do so independently," he states. This is, however seen as a stopgap solution. Some of the interviewees stated that it is SNOMED CT that should be chosen. The majority, however, state that they do not have sufficient knowledge of the details to decide which terminology should be used. Short of developing your own terminology, there is probably no alternative and it would appear natural to choose SNOMED CT. It is stated that not using an existing system as a starting point would require considerably more resources. A number of the interviewees, however, point out the risks associated with the complexity of SNOMED CT and that there are problems with the introduction of such a comprehensive system. "Can we get personnel working on the clinical level to accept/learn this terminology?" was one of the questions. The expectations and needs expressed by the interviewees were to some extent different and the linguistic usage was not clear. The following extracts from the interviews provide a picture of the expectations and needs that exist: Patient safety: for patient information to be shared between patient and care provider and also between different care providers in a manner that is free of risk. "Standardised, common clinical terminology" is needed, primarily for communication in the care chain. The possibility of describing the process in cross-border care between municipal authorities and county councils but also internationally. The primary reason why we need "standardised clinical terminology" is to facilitate the transfer of data between care providers. Co-ordination of the definitions of all terms used in different health record systems would require "uniform clinical terminology". All professional groups must mean the same thing when using the same term. Internationalisation in education and research requires "harmonisation of terminology". The internationalisation of both the health service and the labour market for health service personnel requires the secure transfer of patient information. "International harmonisation" is an inescapable trend. The application of an internationally developed "decision-making support system" presupposes the introduction of SNOMED CT. 19

20 What is SNOMED CT? History For more than 40 years the College of American Pathologists (CAP) has invested in research and the development of SNOMED CT (Systemised Nomenclature of Medicine) [14, 15]. Gradually, SNOMED CT has gone from only including the pathology domain to including a large number of medical areas. In May 2000, SNOMED RT (Reference Terminology) was launched. CAP decided in 1999, together with the Ministry of Health in the UK, to combine SNOMED RT with the UK's "Clinical Terms Version 3" (previously called the READ Codes"). In January 2002, the first version of what is today called SNOMED CT (Clinical Terms) was published. In this version the work that had been done in SNOMED RT within, for example, pathology and medicine, was combined with the work that had been done in the UK, mainly within primary care. The content of SNOMED, calculated in terms of the number of concepts, was doubled with SNOMED CT. Functions for future development and mapping to other systems were also developed. The contents in several domain-specific classifications have been integrated. Examples include English/American classifications for nursing, diseases and health problems and actions. Established classifications (e.g. ICD-9CM, ICD-O) were mapped to SNOMED CT and adjustments in relation to established message standards were made. SNOMED CT has been developed both in terms of content and as a model and can now be described as a concept system developed as a basis for structured care documentation. The purpose of SNOMED CT The value of SNOMED CT is, according to the SNOMED CT organisation, "the potential to support high-quality care of patients and to be able to receive output in the form of health figures for a whole population regardless of the health service system" [15]. SNOMED CT can support the health service organisations with several different health record systems, which must be possible to co-ordinate. It must be possible for co-ordination to take place not only during an individual's lifetime but also between groups of patients and whole populations in order for safe, effective treatment to be given and for diseases to be traced. Without uniform terminology there is a tendency for important development work related to decision-making support and patient safety to proceed far too slowly. 20

21 Facts about SNOMED CT SNOMED CT at present covers a series of activities which provide output in the form of products and services. These describe the SNOMED organisation in general terms as follows: [14] A database comprising three interlinked registers: Concepts, Descriptions and Relations. Tools which support the development and maintenance of the database. Tools which support the presentation of new contents. A service that maintains and develops the concept system. Documentation of technical specifications. Guidelines for introduction. Standards allied to SNOMED CT which are linked to other international information standards. Consultation service for introduction. Education/training service. Concepts The number of concepts in SNOMED CT is approximately 370,000, all of which have their own identity. The concepts are arranged into 18 main categories (see Table 1 below). Examples for each category are given in brackets. Table 1. Main categories (concept hierarchies) in SNOMED CT. Clinical Finding: Finding (Swelling of arm) / Disease (Pneumonia) symptoms, clinical findings and diagnoses. Corresponds to the parts which in care documentation are usually found, for example, under the headings Medical History, Status and Diagnosis Procedure/intervention (Biopsy of lung) Actions and activities that are performed within the health service Observable entity (Tumour stage) Features which, following interpretation or quantification, can be expressed as a clinical finding Body structure (Structure of thyroid) Normal as well as deviating anatomical structures Organism (DNA virus) Micro-organisms, such as bacteria, viruses, fungus etc. Substance (Gastric acid) Chemicals, proteins, liquids, gases etc. Pharmaceutical/biologic product (Tamoxifen) Designations for pharmaceutical substances Specimen (Urine specimen) Systems or components for taking samples, e.g. body liquids, faeces etc. 21

22 Qualifier value (Bilateral) Values for qualifier Physical object (Suture needles) Designations for equipment and materials Physical force (Friction) Physical forces, such as movement, electricity, magnetism, sound, light, heat etc. Events (Flash flood) Events which could cause injury (excluding actions and activities which are listed under Procedure/intervention) Environments/geographical locations (Intensive care unit) Environmental factors and geographical designations Social context (Organ donor) Social and administrative Context-dependent categories (No nausea) Contains "risk of", "negative", "previous history of", "family history" etc., i.e. things which determine the context for another main concept, e.g. "previous history of cardiac infarction" Staging and scales (Nottingham 10-point ADL) Phases and scales Linkage concept - Link assertion (Has aetiology) - Attributes (Finding site) Contains all the relations which occur between concepts in SNOMED CT Special concept (Inactive concept) Special concepts Descriptions Each concept has a descriptive heading. The system also permits synonyms and the total number of descriptive headings with synonyms is approximately 1,000,000, which means that on average there are three synonyms for each concept. For each concept there are at least two descriptions, a recommended term and a fully specified term, which must be unique to the concept. The fully specified term is used to distinguish the concept from other concepts. Relations The number of relations, i.e. named links between concepts within SNOMED CT, is very large, approximately 1.5 million. Table 2 below shows which relations are used to describe concepts from the different main categories Disorder and Finding, Body structure and Procedure. For example, the relation Finding site is used to link the concept angina pectoris to the concept heart structure. 22

23 Table 2. Possible relations in SNOMED CT for the main categories in Disorder and Finding, Body structure and Procedure Disorder and Finding Body Structure Procedure Finding Site Associated With After Causative Agent Due To Associated Morphology Severity Onset Course Episodicity Interprets Has Interpretation Pathological Process Has Definitional Manifestation Occurrence Stage Subject of Information Laterality Part of Procedure Site Direct Indirect Procedure Device Direct Device Indirect Device Procedure Morphology Direct Morphology Indirect Morphology Method Direct Substance Using Access Approach Priority Has Focus Has Intent Recipient Category Access Instrument Revision Status Has Specimen Component Figures 1 and 2 below show examples of two concept descriptions in SNOMED CT, angina pectoris from the category Clinical findings and atrioventricular pacing from the category Procedures. Angina pectoris is described with the aid of the relations IsA, Finding site and Has definitional manifestation, as well as four so-called Qualifiers (Onset, Severity, Episodicity, Courses). The procedure is described with the aid of the relations IsA, Procedure site, Method and Direct device, as well as two Qualifiers (Access, Priority). The model exemplifies how the so-called objectattribute-value trio is used to describe the concept and how the model can act as support for structured care documentation when current values for, for example, Onset, Severity, Episodicity and Courses can be stated for a specific patient. 23

24 Figure 1. Examples of a concept description of angina pectoris from SNOMED CT. To the left is part of the concept hierarchy from the category Clinical finding and to the right how angina pectoris is described with the aid of the relations IsA, Finding site and Has definitional manifestation, as well as four so-called Qualifiers (Onset, Severity, Episodicity, Courses), where the latter can be used for a specification of the concept for a certain patient when one of the alternatives for Onset, Severity, Episodicity and Courses respectively is chosen. Figure 2. Examples of the concept description of atrioventricular sequential pacing from SNOMED CT. To the left is part of the concept hierarchy from the category Procedures and to the right how atrioventricular sequential pacing is described with the aid of the relations IsA, Procedure site, Method and Direct device, as well as two Qualifiers (Access, Priority). Possible areas of application A number of different areas of application of SNOMED CT can be envisaged, although they all need to be tested and evaluated. A possible application is to use SNOMED CT as a basis for structured care documentation. An example of an interface for structured care documentation is to be found in Figure 3 below. For this type of application a number of questions need to be examined. Is the set of relations (qualifiers etc.) for different main categories and concepts in SNOMED CT applicable to care documentation? Do certain qualifiers need to be filtered out or others added? How should the link between concepts in SNOMED CT and an information model for care documentation be handled? Other possible applications are related to the reuse of information for different needs in operational follow-up. Questions for further examination 24

25 deal with how the hierarchies in SNOMED CT can be used for the aggregation of information and to what degree SNOMED CT-based care documentation could facilitate automatic generation of data for health data and quality registers. The potential for decision-making support functions linked to care documentation also needs to be examined further. How would a knowledgebased system with support for individualised care plans be integrated with the next generation of care documentation systems? Figure 3. A screen example from the UK system Clinergy, which was developed to support structured care documentation. The different descriptions: duration, productive character, sputum etc. for the main concept in question 'cough' can be compared with the set of qualifiers from SNOMED CT in Figure 1. 25

26 SNOMED CT in comparison SNOMED CT, the National Board of Health and Welfare's termbank and WHO classifications When assessing terminologies, classifications and concept systems different dimensions have been used [16, 17, 18, 19, 20, 21]. The different dimensions that have been used can be summarised under the headings Purpose, Coverage, Contents Structure and Quality. These dimensions, alongside assessments related to practical use, have also been used in other assessments of SNOMED CT. It should be noted that the different dimensions are interdependent: Purpose: Different purposes have been stated for the different terminologies, classifications and concept systems. The purpose could, for example, be that particular usage within a specific area is employed for follow-up, for input in IT systems or as a reference in conjunction with the exchange of information. These purposes do not need to be exclusory. Coverage and content: Different terminologies, classifications and concept systems cover different areas. The area could, for example, be a medical area of application or one or several professions. The coverage could vary in breadth and depth and coverage could also vary in "density", i.e. how many different levels of breadth and depth are covered. Structure: How different terminologies, classifications and concept systems have been built up. This dimension covers terminological, semantic and technical aspects. It could, for example, involve the handling of synonyms and homonyms, the use of residual classifications, the handling of several specificity levels, the handling of multi-dimensional hierarchies and the "construction" of new concepts (post-coordination). Residual classes (e.g. "other specified Salmonella infections") involve problems as the description of the residual class depends on the other classes in the same section or category. When something has been changed, e.g. when a new class has been added, the significance of the residual class is also changed. In other words, residual classes are not stable over time. Several specificity levels (e.g. infectious disease bacterial infectious disease Salmonella infection (disease) Salmonella pneumonia) are important in order to satisfy differing needs. When documenting, the health service personnel want to be as specific as possible (e.g. the patient has Salmonella pneumonia). When conducting searches and linking to medical knowledge the need is often more general (e.g. does the patient have an infectious disease?). 26

27 To be able at the same time to handle several specificity levels is important if concept systems are to facilitate decision-making support. Multidimensional hierarchies are needed in order to make truer descriptions of reality. The health service is not one-dimensional and consequently the concept systems that are used should not be one-dimensional either. A onedimensional hierarchy requires that a class or a concept only appears in one place in the hierarchy. Salmonella pneumonia must in this case be placed either under pulmonary diseases or under infectious diseases. The user of the terminology or the concept system must then be aware of these choices in order to work efficiently. A multi-dimensional hierarchy does not have this limitation. This is both a significant opportunity, and probably also a prerequisite, to achieve a sufficient degree of coverage [22]. At the same time it is a problem, as meaningless combinations must be eliminated and as it could be possible to describe the same concept in different ways within a concept system. Quality: What is the level of quality within the terminology or concept system and how is this quality assured? Based on these dimensions, different concept systems, terminologies and classifications can be described. As an example, below is a description of the WHO classifications, the National Board of Health and Welfare s termbank and SNOMED CT. WHO classifications The WHO classifications (WHO Family of International Classifications FIC) are used for national and international follow-up and comparison, primarily in causes of death but also to an increasing extent disease panorama. The WHO classifications cover several areas: diseases and health problems (ICD-10) and functioning (ICF). For procedueres there is no common international classification although an abbreviated version (ICHI) of the Australian procedure classification has been proposed as a "reference classification" for health interventions. ICD-10 contains over 12,000 classes of diseases, symptoms and related health problems and covers the need within health service statistics for a diagnosis description on an aggregated level. It does not, however, cover the needs of the health service system on a clinical level [23]. The structure in the classification is one-dimensionally hierarchical, i.e. each class can only have one place in the hierarchy. The classification contains approximately 18 per cent residual classes in order to achieve full coverage. During use, coding takes place on the four- or five-character level although for reporting purposes different degrees of detail can be used, such as using the classification's hierarchical levels: chapter, section, categories and sub-categories. WHO uses a protocol, the WHO FIC Protocol. for quality assurance in conjunction with the adoption of new classifications in FIC. 27

28 The National Board of Health and Welfare's termbank The National Board of Health and Welfare's termbank [24] presents the concepts and terms which are standardised or recommended for use on the local, regional and national level within the health service. The aim is that the different parties within the health service (e.g. principals, authorities, universities and colleges) should use a common special language. The termbank contains approximately 700 defined concepts and terms from different subject fields and is supplemented continuously as new concepts are defined. The National Board of Health and Welfare is responsible for national co-ordination of concepts and terms within the health service and the termbank is a means of bringing standardised and recommended concepts and terms into the public domain. The principal area of use of the termbank is to support human communication, which takes place verbally and in writing. The termbank contains concept diagrams for central concepts. The contents of the termbank are controlled with regard to, among other things, definitions, terms, comments, areas of use and sources. Special quality controls are implemented regularly with external help. National terminology co-ordination takes place ultimately within the Terminology Council, on which the Swedish Association of Local Authorities and Regions and private healthcare providers are represented. SNOMED CT SNOMED CT is a concept system for the health service with the purpose of acting as an international reference for the health service comparisons and aggregation of data. SNOMED CT aims to cover the whole or large parts of the requirements in the health service for concepts and terms in electronic form and to make this sufficient on a detailed level for health service documentation. SNOMED CT is not comprehensive large parts of ICF are missing, for example. For laboratory medicine and pharmaceuticals, external classifications are used, referred from SNOMED CT. SNOMED CT has, despite its extent, demonstrated in studies to have in certain cases a level of coverage of less than 100 per cent (see next chapter). SNOMED CT has a multi-dimensional, hierarchical structure, i.e. the concepts are not limited to one position in the hierarchy. SNOMED CT contains no residual classes as is the case in the classifications, which makes considerable demands on the range of coverage. In SNOMED CT there is often, depending on the area, a large number of specificity levels to choose from. A prerequisite for SNOMED CT is that computer aid is used in conjunction with production, maintenance and, if the potential is to be utilised, also in usage. In SNOMED CT there is the possibility of later on "constructing" new, complex concepts and use is made of computers to ensure the hierarchy concurs logically, i.e. that no contradictions are to be found in the concept system. Despite this, the quality of SNOMED CT has been criticised and several errors in the SNOMED CT concept system have been identified [25]. Shortcomings in quality occur in particular in the fact that parts of SNOMED CT have been modelled incompletely. Generally, it can be said that different parts have been gone through to a varying degree. It is therefore doubtful to 28

29 what extent SNOMED CT can be used as a source of support in making decisions. SNOMED CT, the National Board of Health and Welfare's termbank and the WHO classifications appear to differ on almost every point. They have different purposes, they cover different areas on different levels of detail, they have different structures and quality and they have different ways in which the quality is assured. Information models and concept systems Different types of models are used to describe the health service and its IT systems. Depending on the purpose of the models they can be divided into, for example, information models and concept systems. Information models are used to describe information about specific patients, diseases, activities and so on. The information model is part of the information structure. An information model for blood pressure always contains the patient's identity, the time, the name of the person who made the measurement, how it was done, the results in the form of systolic and diastolic values etc. In concept systems there are descriptions of different types of patients, diseases and activities and they are therefore never linked to a certain patient. A concept system is an abstraction of reality. The concept system can be reproduced in a graphic description of the relationship between concepts, a so-called concept diagram. A SNOMED CT problem One problem, which is by no means unique to SNOMED CT, is that in practice there is no clear division between what is described in a certain information model and what is described in a certain concept system. This problem has come to the fore, particularly in the United Kingdom, with the concurrent introduction of SNOMED CT (a concept system) and HL7 2 RIM (an information model). There are, for example, several overlaps between SNOMED CT and HL7 RIM. In both models there is a means of describing negation, i.e. that something does not exist. In HL7 RIM there is the attribute "mood-code", which is used to describe the status of an activity if the activity has occurred and if it is an order, an objective or a precondition. The same functions are also handled, albeit differently, in SNOMED CT. There are, in other words, several ways of describing the same thing in the combination SNOMED CT + HL7 RIM depending on where one practically (and arbitrarily) draws the line. Finding general solutions to this problem has proved to be very difficult. There is a special working group within HL7 and CAP/SNOMED CT called "HL7 Terminfo", which works with interface issues between HL7 and SNOMED CT and to provide recommendations regarding the use of SNOMED CT concepts within the area of the HL7 standards. A draft of the implementation guide has been produced, which addresses some of the problems identified [26]. 2 American supplier standard for health service information systems. 29

30 Evaluations and applications Scientific studies of the content of SNOMED CT Studies of earlier SNOMED CT versions show that SNOMED CT was the concept system which in comparison with classifications and terminologies had the widest level of coverage with regard to overall health record contents [27, 28]. As the content of SNOMED CT has been expanded there is reason to believe that these results still hold today. Examinations of the new SNOMED CT show similar results for more general health record information [29, 30] and also for its specialist areas: ophthalmology and "Guidelines for Evaluation and Management of Chronic Heart Failure" [31]. The latter study compared the degree of coverage with seven other classifications and demonstrated that SNOMED CT covered 100 per cent of the concepts within seven of 14 different domains and had better coverage than any of the classifications. A new study at the Mayo Clinic shows that 4,996 terms from clinical operations which describe health problems could be represented in SNOMED CT up to 92.3% [32]. Another example is the USA Food and Drug Administration's (FDA's) list of indications for oncological pharmaceuticals, made up of several composite concepts which are not represented in SNOMED CT [33]. There are other areas where SNOMED CT lacks good coverage, such as social welfare. Some domains, such as occupational therapy and physiotherapy, are highlighted insufficiently in the research. As regards the nurses' demands on the content of SNOMED CT, there is an active "nursing working group" where leading nursing researchers have taken part in the development work [34]. Evaluations in Sweden The REFTERM project [15] was a collaborative venture between the Karolinska Institute, Stockholm County Council and Stockholm Municipal Authority, the aim of which was to evaluate the possibility of establishing a Swedish reference database for medical terminology based on the international work within SNOMED CT. In one of the sub-projects, interviews were conducted with three doctors from different specialities to pick up views on the SNOMED CT section compared with corresponding codes in ICD-10. The results from this sub-project showed, among other things, that there appears to be a wish for a higher level of detail and more detail levels in the clinical context than what can be satisfied by ICD-10. SNOMED CT as a whole is considered to have the potential to satisfy several needs among doctors in clinical operations if it is translated and for research and followup without translation being required [35]. In another sub-project [36] map- 30

31 ping was conducted between free text in ten patient health records to SNOMED CT concepts. The results showed that the concepts in the health records which were used in the study were relatively easy to classify and code to SNOMED CT and that there is reason to examine SNOMED CT in clinical use, research and follow-up with regard, among other things, to user-friendliness, reliability in coding and functionality of tools for data search from larger patient databases. A third study of 66 concepts for procedures within home-based healthcare indicated that 72% of these had coverage of non-composite SNOMED CT concepts. In addition, the existence of subordinate and superordinate concepts was examined. The study showed that SNOMED CT was quite easy to map to concepts for care planning in advanced home-based healthcare and that 30% of the actions lacked subordinate concepts [37]. Evaluations, international Scotland In Scotland, an evaluation was made of the content of SNOMED CT and its usability for clinical purposes. Representatives from different specialities and professional groups examined the content of SNOMED CT. The report supported the lack of coverage of terminology for physiotherapists, occupational therapists, dieticians, nurses and for the social services field. Certain specialities within emergency care were also considered to lack concepts even if it was felt to a greater degree that SNOMED CT was an improvement compared with existing terminologies. GPs had no major comments and this was expected as the READ codes, which were developed within the NHS over a period of 20 years for use in general practice, are included in SNOMED CT. Clinical usability was also discussed from the point of view of searchability and integration with health record systems. Costly adaptations in existing systems to achieve usability were predicted [38]. The conclusion of the report was that SNOMED CT did not at the present time satisfy the requirements for a joint concept system within the NHS but that it had potential to be a useful tool in the development of electronic health records. This evaluation is one of few, perhaps the only one, that address terminology for social services, physiotherapy, occupational therapy and dieticians. The evaluation took the form of a project report and was not made using scientific methods. The report itself highlights certain shortcomings in the method, such as the fact that several specialities were lacking and that only a few people evaluated certain areas. There are no later studies available to support the assertion that the development of SNOMED CT took place in such a way that action had been taken to counteract the shortcomings which the report highlights with regard to the content for physiotherapy, occupational therapy, the dietician field and social services. Australia An evaluation project in Australia called "GP Vocabulary Project", produced GP terminology during phase 1. During phase 2, an evaluation was made of the classifications DOCL, CATCH, ICD-10-AM, ICPC-2-Plus and 31

32 the concept system SNOMED CT against the GP terminology that had been produced [39]. ICD-10-AM is the Australian modification of ICD-10. ICPC-2-Plus is the expanded Australian adaptation of WONCA's ICPC-2 (diagnoses, procedures and reason for encounter in primary care). (WONCA is the World Organisation of Family Doctors). CATCH is a classification developed to support clinical and service-related documentation on the municipal level. DOCLE is a classification system developed in Australia. Each classification is evaluated carefully and they are also compared with each other. The evaluation covers the following parameters: Well-adapted hierarchies and concept definitions Stability Compliance with terminology for GPs Analysis of the reliability, validity and correctness of results Difficulties, e.g. technical problems which arose in the evaluation process Knowledge of the different terminologies and classifications in relation to the intended use within general medicine Design: id, terms, hierarchies, relationship roles, mapping, support functions, delimited data Contents: size, focus, details, complexity, development methods, including IT systems, human resources and validation methods Relevance to GPs and the health service in general. Current use and contents Administration: organisation, data maintenance, updates Costs and availability Usability for programmers and end-users SNOMED CT is considered to contain more terms and concepts than any other system. The statement about SNOMED CT was: "Its 'polyhierarchy' and types of relations are many and complex. More extensive tests in health record and decision support systems remain. It is probable that SNOMED CT will be an international concept system for the health service. Its most significant strengths are its architecture, focus and detailed content as well as its international support/management. Its weaknesses are mainly 'noncomposite concepts (pre-coordination), non-intuitive, incomplete and defining relations and (for certain users) the disruptions that could arise from undesired concepts. All these weaknesses can be overcome." SNOMED CT was also compared with the Australian classifications in a points system, where the different variables that are evaluated were given scores and weighted in relation to two parameters: its usability in general medicine and its usability in the whole health service system. In this comparison, SNOMED CT received the highest number of points: 32.2 (general medicine) and 31.6 (health service) compared with ICD-10-AM: 7.1 and 6.6 and ICPC-2-Plus: 7.1 and

33 Denmark In the Danish "Sundterm pilot" project, SNOMED CT was translated and validated for Danish use. The comparisons were made between the classifications currently being used in Denmark and SNOMED CT in the view of the need for an international reference for the health service concept systems, linked to the basic structure of the electronic health record (GEPJ) produced by the Danish National Board of Health. The requirements that were evaluated were: the existence of clinical terms and expressions (including synonyms), support for the logic in the clinical process, the opportunity to find a relevant term without difficulty, a flexible level of detail, interprofessionality and a broad level of acceptance. The conclusion that can be drawn following evaluations was that "SNOMED CT is the only product that satisfies the stated requirements for a clinical concept system despite the fact that SNOMED CT at the present time only exists to a limited extent in Danish". [40]. Applications SNOMED CT is relatively new and is not used broadly throughout the world. In the USA, there are a number of large organisations that include SNOMED CT in their system. In Europe, it is the United Kingdom and Denmark which on a national level are investing most in the introduction of SNOMED CT. In Sweden, there is no practical application of SNOMED CT. There is a considerable lack of usable and published references dealing with practical use and which could act as guidance in the introduction of SNOMED CT. Nor is the documentation of SNOMED CT's basic structure, user manuals and implementation guides openly available. USA The Department of Health & Human Services (HHS) reached agreement in 2003 with the College of American Pathologists (CAP), which means that SNOMED CT has become freely available in the USA. SNOMED CT is also included to a large extent in the National Library of Medicines' Unified Medical Language System (UMLS ). A number of agreements are being drawn up between CAP, the owner of SNOMED CT, and suppliers of IT systems in the health service although it is difficult to gain a clear picture of the extent to which the suppliers use SNOMED CT in their production. Kaiser Permanente, a large non-profit company in the USA in the healthcare sector, uses SNOMED CT as an important source in its common terminology service CMT, the contents of which serve all applications within Kaiser Permanente. CMT provides concept definitions and codes in a uniform way within the organisation. Their experiences are presented in a scientific publication [41]. Veterans Affairs, a major healthcare provider, also uses SNOMED CT in its task of creating groupwide terminology for health record systems [42]. The number of health service systems is 128 out of a total of 1,300 units. Kaiser Permanente and Veterans Affairs have in partnership produced a "problem list" with 12,000 concepts from SNOMED CT. The Food and 33

34 Drug Administration (FDA) in the USA has recently adopted SNOMED CT as a standard to present information in conjunction with the prescription of drugs and refers to the aforementioned "problem list" from Kaiser Permanente and Veterans Affairs [43]. Cerner, a global company in the healthcare sector, states that it has been using SNOMED CT since 2002 and has an agreement which permits unlimited use of SNOMED CT in its applications [44]. Examples of applications are care plans for nursing based on SNOMED CT. Another example is the use by Los Angeles County Public Health of SNOMED CT and two other standards to achieve structured laboratory information from local laboratories with the aim of discovering bioterrorism [45]. Denmark In the Sundterm project at the Danish National Board of Health, parts of SNOMED CT are being translated and validated for Danish use. The project will create a decision-making basis to produce Danish health service terminology that can cover the documentation requirements of the clinics in a structured electronic health record and which can provide cohesive health service terminology of both a general and specific nature. The project translates 10,000 concepts per month and 60,000 concepts (20 per cent) had been translated in February A major project in the Copenhagen area, "Health Service Content" (SFI) in the electronic health records, is collaborating with the National Board of Health. "The Classification Project" is also working regionally in Copenhagen and the surrounding areas on harmonisation of terminology and classifications. The project in Denmark highlights the challenge of co-ordinating several local projects, both with each other and on the national level, in order to achieve a uniform information structure. One example is that the translated SNOMED CT version only contains one term per concept, i.e. that the synonyms which are often found in the English version of SNOMED CT are not, on the recommendation of CAP, translated. Synonyms will be added later in conjunction with clinical use. The absence of synonyms means that searchability in the terminology is worsened and that there ought to be a project that co-ordinates local use of synonyms with the national terminology. Other examples are that the concepts which are lacking for local use ought to be complemented in a structured way, either in the national or in the local terminology, and that it must be possible for a terminology service to deal with the handling of versions of terminologies on different levels (national, regional, local). Other questions from Denmark are the extent to which SNOMED CT's structure can be of benefit in building up similar structures for terminologies and classifications and when it is appropriate to use a terminology in conjunction with care planning and programme work. In Denmark, this is done after the care plan groups have concluded their work and created their own terms for activities and clinical findings. Another county council, in addition to the Copenhagen region, has decided to work according to the principle of "health term content" in its 34

35 health record project. There is, however, no political decision in Denmark on the introduction of national terminology in the health service based on SNOMED CT. United Kingdom "SNOMED: a distraction in the short-term and an advantage in the longterm?" SNOMED CT is used within the current National Health Service (NHS) project "Connecting for Health". There is a realisation that the use of SNOMED CT will reduce the risk of incorrect interpretation of information and deviations which arise in conjunction with the use of paper-based health records. SNOMED CT is updated by a central authority and there is possibilities on the local level to request supplements. One example of central use is "The dictionary of medicines + devices (dm+d)" which provides, among other things, medicines with codes and terms linked to SNOMED CT concepts. In the short and medium term, the NHS does not believe that SNOMED CT will replace current classifications. As introduction has commenced, the NHS and its suppliers have also begun to acquire a great deal of experience. As regards decision-making support for pharmaceuticals, the supplier responsible does not consider that SNOMED CT was created for this task but it could be used as terminology at the interface and as a kind of "encyclopaedia". The specific information required for certain information support must be added it cannot be expected that SNOMED CT contains every detail and its own ontological relations must be created for each form of decision support. Experience of the introduction in existing health records indicates that it is something that must be done with a shoehorn and which can be solved but in some cases requires considerable changes in the applications. There is reason to follow closely the introduction of SNOMED CT in the NHS and learn from their experience [46]. 35

36 SNOMED CT ownership and management The organisation behind SNOMED CT SNOMED International is the department at the College of American Pathologists that today owns and organises work on SNOMED CT. SNOMED International [47] is a multidisciplinary organisation with members from CAP, experts on the clinical content, experts in medical informatics and representatives from NHS Connecting for Health (NHS CfH), who work in close collaboration with representatives from different clinical specialities, authorities, doctors and nurses, medical translators, editors, evaluators and external parties. Within CAP it is the SNOMED International Authority that has direct responsibility for management and control of all activities related to terminology. The SNOMED International Editorial Board is responsible for scientific directives, editorial work and scientific evaluations. SNOMED Working Groups vary in number and size and have different levels of detail in their work and in the contribution they make to development. Offer of participation in SSDO NHS CfH and CAP have together proposed setting up a new organisation for the development of SNOMED CT. In an invitation to over 100 countries it was proposed that a SNOMED CT Standards Development Organisation (SSDO) be set up. If the new organisational model is supported by a sufficient number of countries, all processes which to date take place within SNOMED International will move from the present management to the new organisation. Facts about SSDO The brochure sent out with the invitation includes a description of the vision of a common SNOMED organisation: that several countries will have the opportunity to own and control one of the key elements in all electronic health records, i.e. the concept system which both computers and clinics need and use in the health service. It is stated that SSDO would ensure highquality development of an extensive international reference for health service concept systems. The vision for the SSDO organisation is described in the brochure as follows: "SNOMED CT is the information standard that will support international interoperability and agreement with regard to the care of patients in the 36

37 health service organisations which are supported by effective IT systems. The management process within SSDO will ensure that collaboration, alliance and capacity for appropriate location of the work are supported." It is envisaged that SSDO and national centres will share a common infrastructure and co-operate in real time. National centres will have the right to self-determination and can themselves control the level of influence SSDO will have in local development. They can also co-operate between themselves in different constellations. SSDO will be based on three basic principles: Clinical purpose SNOMED CT is a concept that is being developed by clinics, primarily to support health service personnel in the care of individual patients. Clinical integrity SSDO will describe and develop necessary processes to deal with concepts in such a way that the clinical and technical integrity in the product is assured. Economic viability SSDO will be financed as a non-profit company in a way that provides the first two basic principles and assures the longterm viability of the concept system as a bearer of lifelong health record information. In the information brochure that accompanied the invitation [47] to participate in SSDO, there is a description of different types of membership. The information is summarised below. Membership Charter membership The countries that join SSDO initially receive a special invitation which involves greater participation in the management and development of the organisation. The initial invitation is directed at the highest political levels in the countries. These will own, manage and control SSDO. Full membership Full membership will be offered to national organisations that are to a large extent financed through state funding. These organisations can become members of SSDO after its formation. (This alternative applies to Sweden if it joins as a country.) Affiliate membership Individuals and organisations that do not have the status required for full membership can still be affiliated to SSDO through this form of membership. Vendor membership Suppliers that develop software or systems and aim to distribute applications that facilitate the use of SNOMED CT can also become members of SSDO. Legal framework The business model presented in the information brochure only presents overall costs. The starting point, however, is that no organisation shall make 37

38 a financial profit. It is therefore proposed that the financial framework in SSDO shall be such that it supports a non-profit organisation, NPO. Organisation Management The Management Board at SSDO will be responsible for: All decisions within SSDO All aspects of the organisation structure, processes, outputs, management, viable financing, long-term solvency, legal position, quality and integrity in the SNOMED standards The maintenance of the vision as well as basic principles within SSDO. The Management Board shall comprise 12 voting members as well as a chairman. The members will represent four geographical regions Europe, North and South America, Asia/Australia/Oceania and Africa and the Middle East. The representatives from each region should include a clinician, a manager/political representative and a clinical terminologist. Harmonisation Boards Four Harmonisation Boards will work with the Management Board Terminological Content, Technical Structure and Infrastructure, Business Operations and Research and Innovation. Each committee will be led by a board member. Working Groups The Working Groups will support the Harmonisation Boards. The groups will not have a formal structure but will remain flexible in order to satisfy different needs. They are initiated and set up by the committees and they have a task-oriented composition of relevant experts. 38