Terminology & Classifications News

SNOMED CT Code Usage Data

Published: April 2022

Dear colleague,

NHS Digital has now published the SNOMED CT code usage data for 2019-20 and 2020-21. It can be found at SNOMED Code Usage in Primary Care - NHS Digital.

This published data was aggregated from more than one original mandatory data return provided by individual system suppliers under the GP IT Futures programme, as detailed here: Primary Care Clinical Terminology Usage Report - GP IT Futures Capabilities & Standards

  • Each annual report lists all SNOMED CT codes from within the SNOMED International core or the UK's Extensions PROVIDED at least one new EPR entry had been made using the code
    • SNOMED codes for medications used (dm+d codes) are however mostly absent, because the original data extracts from contributing supplier systems do not include the prescribing record
    • SNOMED codes received as part of data flows from laboratories sending results are included
  • All counts are rounded to the nearest 10 (i.e., an original aggregate value within the range 95-104 is published as 100), except the very small non-zero counts of 1,2,3 or 4 that are instead rendered as an asterisk
    • SNOMED codes that achieved zero usage during the reporting period are not listed at all, even if they were part of the prevailing release of the UK Edition of SNOMED during the reporting period.
  • The published counts are NOT pre-standardised to a constant population. Such standardisation across annual reports would be possible by reference to the co-published metadata report, which states the number of patient records from which the counts were likely to have been aggregated (see Primary Care Clinical Terminology Usage Report - GP IT Futures Capabilities & Standards for more information on what that metadata patient count is actually reporting)
  • Two additional columns report whether or not each listed code was "active" within the UK Edition of SNOMED CT on the first and last days of the reporting period

The metadata file lists the significant caveats, primarily that these are of course counts only of how often clinicians choose to record something. There is therefore only a very weak and unstable correlation between how often codes for clinical phenomena are recorded and the true incidence or prevalence in the real world of those same underlying clinical phenomenon. Some clinical phenomena never get coded and others only some of the time. Some conditions may be recorded more than once for the same patient during the same reporting period. The data also displays clear evidence of miscoding (wrong code selected) and undercoding (code selected is true but underspecified).

Kind regards

Terminology & Classifications Delivery Service

Read all news articles

Created by Katy Park 2 months ago