Background/Purpose: Clinicians use a range of different terms to describe the same clinical concept. Whilst these variations seldom lead to confusion among clinicians they make it difficult to aggregate, analyse and share clinical data. The ability to create and share large data sets has become increasingly important in the age of personalised medicine when it is not possible to collect enough data from a single centre to tease out the importance of phenotypic variations. Similar restrictions apply when studying geographical and temporal trends in case mix.

Methods: Working with the British Society for Rheumatology (BSR) the authors have developed a list of diagnostic terms (standard term set (STS)). The BSR Specialised Adult Rheumatology Specification was used as template. The principles underlying the structure of the STS are that it should reflect anticipated use cases for the collected data, the set should serve equally the needs of general rheumatology departments & highly specialised units, the set should allow broad concepts and more granular ones to be collected with equal ease and that it should be mapped to ICD 10 and SNOMED CT as fully as those instruments permit. A hierarchical structure to the STS allowed aggregation of many concepts to a single parent while retaining the ability to separate out conditions of particular interest. The STS will form the core of a much richer data model for RMDs which will include minimum data sets for each concept in the STS and capture comorbidities, phenotypic variables and outcome measures. In this phase the authors have made no attempt to capture diagnostic uncertainty, detailed phenotype, secondary diagnosis or co-morbidities. Where there was a need to reflect new knowledge or new classification of diseases the authors have also undertaken a programme of SNOMED CT content improvement in collaboration with the UK Health and Social Care Information Centre (HSCIC) to improve representation of current concepts and ring fence it for ease of future use. The authors also developed a synonym table which maps physicians combined clinical vocabulary to the matching STS. The final set for the first phase of this coding project was arrived at after a number of iterative editorial cycles with valuable input from local musculoskeletal coding leads for ICD 10, HSCIC and lead clinicians from pilot sites. The STS was released to pilot sites in early 2016 along with a detailed cover note, reference guide for users, agreed “minimal data set” and hyperlinks to ICD 10 and SNOMED CT browsers. Minimum data set in this pilot phase comprised of patient identifiers, demographics, appointment date and type, diagnosis and outcome.

Results: The first 3 months of the pilot have shown that large volumes of data can be collected and shared using the STS with minimal additional workload for the clinician. It has been successfully implemented in various clinic settings ie community, district general hospital and large teaching hospitals. It serves the purpose of both general & special interest clinics. The STS can be used to capture data using paper, electronic & digital care records. It can be used by a single user or entire department. For the first time it has allowed clinical data collected during routine outpatient consultations to be used for a variety of secondary purposes including case-mix analysis, commissioning, workforce planning, audit and identification of research cohorts. Data collection is ongoing. Results will be presented at the ACR meeting.

Conclusion: Clinicians should lead the development of health informatics solutions. Software developers should work closely with health professionals to build platforms that collect data as part of routine care using models that place data in context so as to maximise secondary use ie data collected once can be used with equal ease to provide care for an individual or study trends at a population level.

ACR Meeting Abstracts - https://acrabstracts.org/abstract/clinician-led-development-of-a-standardised-term-set-for-rheumatic-and-musculoskeletal-disorders-allows-easy-creation-of-large-scale-icd-10-and-snomed-ct-mapped-datasets-from-routinely-collected-clini/