The success of an implementation of SNOMED CT is often driven by the quality and approach that has been employed in the development of the user interface. This is where end users will interact and use SNOMED CT and a poor implementation will have a considerable impact on the quality of data entered. These points provide some guidance on what should be expected from vendors in systems using SNOMED CT.

1. Use SNOMED CT for searching, display, storage, communication, knowledge linkage, querying and analytics in the following clinical domains (or specific data elements)
   • Problems, Diagnosis and Clinical Findings
   • Reason for Admission
   • Procedures (e.g. Planned Procedures, Performed Procedures)
   • Allergies
2. Support searching for SNOMED CT concepts using any term that is preferred or acceptable in the national language reference set.
3. The vendor shall include one of the following options for each SNOMED CT coded data element (depending on user preferences and local requirements for standardization of interface terms):
   • Upon selection of a concept, use the term entered by the user to display the selected concept; OR
   • Upon selection of a concept, use the preferred term from the national language reference set) to display the selected concept; OR
   • Upon selection of a concept, use the preferred term from the regional or institution specific language reference set to display the selected concept.
4. The vendor shall allow searching and selection of only those concepts from the SNOMED CT subset that has been specifically bound to that data element.
5. The vendor shall display the most frequently selected concepts at the top of the list, for elements bound to a subset containing more than 20 concepts.
6. As the user types each character into a SNOMED CT coded data element, limit the selection of concepts to those with a preferred or acceptable term that matches the characters types (using a ‘word prefix any order’ algorithm), and use auto-complete when only one option is available for selection.
7. When displaying the list of possible matches, display the concept with the shortest matching term first.
8. For each free text data element that records clinical information (e.g. Past history, Clinical notes) use SNOMED CT-enabled techniques (e.g. Natural Language Processing) to suggest possible SNOMED CT encoding (including appropriate contextual information), for selection and confirmation by the user.
9. Support the capture of SNOMED CT post-coordinated expressions using predefined expression templates and automatically-generated interface terms – for laterality, allergies and family history.
10. Use SNOMED CT concept identifiers stored in the EHR to suggest patient-specific clinical knowledge to the clinician, and to test clinical decision support rules.
11. Use SNOMED CT codes stored in the EHR, together with SNOMED CT and other map-enabled coding systems (e.g. SNOMED to ICD-10) to suggest appropriate codes for the clinician to select (if required).
12. The system shall be sufficiently performant to return the first (X) results in a given time period (e.g. less than 1s).