Various studies have described methods for the development of terminological subsets [25, 27, 33,34,35]. Most of the studies referred to and used a process model (or aspects of one), meaning that a process from creation through to maintenance was described. The conclusion can be drawn that a process model seems to be promising as a method for developing a nursing subset. However, there also seems to be a lack of uniformity in the stages, approaches and techniques, so the process models have not yet been fully explored and are still evolving [36]. In this study we used the process model for the development and maintenance of subsets as part of the International Release of SNOMED CT as described by the IHTSDO [37] and the Dutch instruction ‘making a SNOMED CT subset’ derived from it and set up by Nictiz [38], which involved the following six stages: 1) Scope/Requirements; 2) Design/Planning; 3) Development; 4) Distribution 5) Implement and Use; 6) Maintenance. Figure 2 gives an overview of the stages. Each stage will be explained in the next paragraphs.
Stage 1: Scope/requirements
In this stage, we defined the purpose of the subset and relevant requirements, such as the scope of content and the users. First an expert team was set up, consisting of a researcher (RK) and a nursing expert (EV), both with extensive knowledge of the structure and content of SNOMED CT, and two representatives of Nictiz (acting as the Dutch SNOMED CT Release Centre): the Terminologies Coordinator (PV) and a terminologist (EG).
The expert team identified the scope, which was to develop a national nursing subset of patient problems based on the SNOMED CT terminology to assist interoperability. The users of the subset were defined as clinical nurses working in various healthcare settings. Other existing subsets were then explored to evaluate whether they met the requirements. To our knowledge, two national SNOMED CT nursing subsets of patient problems have been developed, namely a United States (US) [25] and a Danish [26] nursing subset. Denmark developed a national homecare nursing subset of 80 concepts (not available online yet) building upon the US nursing subset [26]. Patient problems from the US nursing subset were retrieved from the ‘Unified Medical Language System’ (UMLS) Metathesaurus database, developed by the National Library of Medicine (NLM) [25]. The database contains concepts from various different classifications and terminologies. Queries were performed by the UMLS to collect patient problems from SNOMED CT and four nursing classification systems (the Omaha System, NANDA International, the Home Healthcare Classification (HHC) and the ICNP). The patient problems included were reviewed manually and discussed, resulting in 369 nursing problem concepts (https://www.nlm.nih.gov/research/umls/Snomed/nursing_problemlist_subset.html). Although this subset could be useful for building on, we decided to develop a new subset. The main reason for this decision was the findings of a previous study, ‘A nationwide survey of patient problem occurrence across different nursing healthcare sectors’, in which Dutch clinical nurses were asked to indicate which patient problems they encountered most frequently in daily practice, as well as the influence nurses said they had on these problems [28]. This resulted in an overview of patient problems (version 0.1) reflecting the Dutch clinical nursing practice across healthcare settings. Using this overview as a framework we could specify the patient problems as identified by nurses themselves. This approach differs from the US subset, which contains concepts from various different classifications and terminologies (regardless of their occurrence or the perceived level of influence).
Stage 2: Design/planning
In this stage, we defined the composition, the involvement of participants, sampling and recruitment process. Focus groups with nursing professionals were held to determine which SNOMED CT concepts cover the patient problems (version 0.1). In order to recruit participants, an invitation to participate in the focus groups was sent in a digital newsletter from the Dutch Nurses’ Association (V&VN). This monthly newsletter was mailed to 70,000 members of the Dutch Nurses Association, giving information about the study as well as the registration process.
Sixty-seven nurses replied to the recruitment message in the newsletter and agreed to participate. We organised seven focus groups, using the following inclusion criteria:
□ Employed as nursing professional.
□ At least 2 years’ experience as a nursing professional.
□ Working in hospital care, residential care, psychiatric care, primary care or care for the mentally disabled.
Because the nurses were active in a variety of nursing practice contexts, patient problems could be discussed from different perspectives, which was necessary to determine whether the patient problems were comprehensive, unambiguous and acceptable in a broad nursing context. The expert team also took part in each focus group.
The focus group meetings lasted two and a half hours. The nursing expert (EV) led the meeting, explained the procedures, and introduced the method and the patient problems to be discussed. The terminologist (EG) identified and selected corresponding SNOMED CT concepts and ensured that the concepts were consistently and accurately applied in line with the SNOMED CT guidelines. The nursing researcher (RK) observed and monitored the process.
Stage 3: Development
There is a variety of approaches for developing subsets, such as developing a new reference set or adopting, copying and adapting an existing reference set [37, 38]. In this study, developing a new subset was deemed appropriate, firstly because the development could build upon the existing overview of the study mentioned earlier in which 440 Dutch nurses had already participated [28] and secondly because the involvement of nurses could be maintained in order to improve backing and approval of the final subset.
The development process was set up in four phases [38]: 1) the selection of SNOMED CT concepts; 2) review and translation process with focus groups; 3) defining and modelling; 4) validation of the subset. This setup was based on the Dutch Nictiz instruction ‘Making a subset’ [38].
a) Selection of SNOMED CT concepts
The first phase comprised selection of SNOMED CT concepts by the expert team. The overview of patient problems (version 0.1) acted as a framework (Additional file 1). The patient problems (version 0.1) contained both Dutch and English terms. The expert team then selected and identified a matching SNOMED CT concept (or the nearest match) for each patient problem based on the term and definition in version 0.1. The concepts were selected from the core distribution of the International SNOMED CT Edition (January 2016 release) managed by SNOMED International and available online at http://browser.ihtsdotools.org/.
An example of the concept ‘Pressure ulcer’ from the core distribution of SNOMED CT is shown in Fig. 3. The concept has a unique numeric identifier (399912005) and equivalent synonyms (Contact ulcer, Pressure sore). Each concept is linked to a more general concept in the hierarchical structure, the so-called ‘parent’. In the example of a ‘Pressure ulcer’, the parent is ‘Chronic ulcer of the skin’. It is also possible to specify ‘Pressure ulcer’ in increasing detail. The specifications are referred to in the underlying hierarchy as ‘children’, for example ‘Pressure ulcer stage 1 and stage 2’ and so forth.
SNOMED CT concepts that were equivalent to concepts from the International Classification of Nursing Practice (ICNP) were preferred in order to ensure that the terms accurately represented the nursing domain. The ICNP is a formal terminology for nursing practice developed by the International Council of Nurses (ICN) [39]. SNOMED International and the ICN collaborated in order to harmonise both terminologies to increase interoperability and to encourage the use of terms as established by the ICNP [40]. SNOMED International and ICN developed an ICNP-to-SNOMED CT Equivalency Table for Diagnosis and Outcome Statements [41], meaning that each ICNP diagnosis included in the equivalency table has the same meaning as the SNOMED CT patient problems included (English edition, release version 20,160,131). The equivalency table was used to ensure that the SNOMED CT concepts matched consistently.
b) Review and translation process (with focus groups)
In the second phase, the patient problems plus matching pre-selected SNOMED CT concepts were reviewed and discussed. Each focus group discussed and reviewed an average of 12 patient problems. Both the patient problem from version 0.1 and the matching SNOMED CT concept were presented to the participants of each focus group. The SNOMED CT concepts were presented directly from the browser (see the example in Fig. 2) so that the hierarchy could be clarified by switching between different concepts and their parents or children if necessary. The participants discussed the preselected concepts using the following predefined questions:
-
Is the term sufficiently comprehensive for electronic recording?
-
Is the term unambiguous and understandable?
-
Is the term professionally acceptable for nursing practice?
These questions were derived from the viewpoint of the Nursing Special Interest Group on the nursing contribution to quality assurance of SNOMED CT [42]. Nursing professionals participate in the Nursing Special Interest Group to advise IHTSDO on ‘the development, validation, uptake and implementation of SNOMED CT and related products’ [43] (p. 4).
Each concept had a SNOMED CT term derived from the English edition (release version 20,160,131). The terms were translated to Dutch following the SNOMED CT guidelines for translation [44]. The nursing professionals from the focus groups and the expert team were involved in the translation process. Nursing professionals confirmed that the preferred Dutch terms corresponded to the terms used in their daily activities and were clinically acceptable.
The SNOMED CT patient problems included in the equivalency table have the same meaning as the ICNP diagnosis. We were therefore able to validate the translation process by using the Dutch catalogue from the International Classification of Nursing Practice (ICNP) [45]. The ICNP beta version, including terms and definitions, was translated (working in both directions) into Dutch by the Dutch Nursing Union (Nu’91) in cooperation with the ICN.
Once a focus group reached a consensus about a concept, the terminologist coded the selected concept. If a focus group did not reach a consensus about a concept, it was debated in another focus group until a consensus was obtained. If the groups found a concept to be either inconsistent or incomplete or if there were no appropriate concepts, requests for additions or changes to SNOMED CT or new concepts for it were submitted to the Dutch National Release Centre (Nictiz).
c) Defining and modelling
In the third phase, the expert team defined each SNOMED CT concept in Dutch (in SNOMED CT terms: ‘textually defining’). The (textual) definitions provide additional information about the intended meaning or usage of each concept. To ensure that the meanings of nursing concepts were reflected accurately, national Dutch guidelines were examined and the definitions available in them were used where possible. If no definition was available, the definitions of nursing diagnosis as established by the International Classification of Nursing Practice (ICNP) were used; these were also described in the Dutch ICNP catalogue [45]. If no definition was available in the ICNP catalogue, the definition from another classification was used (for instance the International Classification of Functioning and Disability).
After each focus group, the expert team broke the selected SNOMED CT concepts down into two items, a name and a textual definition. A SNOMED CT concept could be expressed as a single clinical finding or as a judgement about a focus (as described in the “Conceptual framework”). The terminologist also ensured that the concepts were consistently applied and accurately coded in line with the SNOMED CT guidelines [46, 47].
The participants in each focus group were given an overview of the terms and (textual) definitions discussed in their meetings to review as a final check.
d) Validation of the subset
The final subset, consisting of SNOMED CT patient problems with corresponding terms and definitions (n = 119) and associated SNOMED CT codes, was presented to all participants (n = 67) to determine if nursing practice was consistently covered. All the participants also confirmed that the terms and definitions accurately reflected nursing practice and that the terms used were unambiguous and understandable.
The nursing subset of SNOMED CT patient problems was also presented to the SNOMED International Nursing Special Interest Group, who were asked to review it to ensure consistency.
Review of the subset needs to be maintained over time, both to review the subset against specified use cases and to accommodate changes to existing content or add new SNOMED CT content. Separate review projects are being set up, but were beyond the scope of this study.
The final subset was distributed in an electronic format and released online. Each patient problem includes a link to a common feedback form where nurses are encouraged to make recommendations or request revisions, additions or new concepts.
Stage 4: Distribution
Subsets can be distributed as part of the International Release, as part of a National Edition or as part of an Affiliate Edition [37]. For this study, it was decided that the subset will be distributed six-monthly as part of a National Edition, which is in line with the distribution frequency of the International Release. The standard format for distributing the SNOMED CT subset is a Simple Reference Set representing an extensional definition of a subset of components (more information about a simple reference set type can be found in the SNOMED International Practical Guide to Reference Sets [37].
Stage 5: Implement and use
When a subset has been developed, it should be implemented for use in nursing practice. Implementation means that the subset should be integrated into software systems. It is important to support the implementation with guidance during implementation. Additionally, collaboration with users and vendors is necessary in order to test the intended use and its effectiveness. The implementation in software systems and use in practice were not included in the scope of this study and will be followed up with another study.
Stage 6: Maintenance
This stage consisted of establishing a management and maintenance structure, including change management and the revision cycle. The management and maintenance structure was set up in line with NEN 7522:2010 nl ‘Maintenance of coding systems and other terminological systems’, which is a standard defining roles and responsibilities of organisations and people involved in the development of terminological systems. It is applicable only to Dutch healthcare [48].