Clinical assurance has been sought from both medical and pharmaceutical professionals throughout the process of identifying patients for inclusion in the SPL.
The approach taken built in a number of safety checks. These included the principles to:
- reuse code lists which are already established in the system
- minimise burden to the system including GPs and IT suppliers
- keep to existing data flow pathways wherever possible to minimise risk
Existing data sets do not hold data in the form in which it was required for this purpose. Data is held in clinical codes which do not directly map to the requirements in the SPL.
Expert clinicians were asked therefore to ‘translate’ (or map) the clinical requirements of the list into the right subsets of coded information so that individual patients could be identified.
The clinical assurance undertaken was restricted to the translation of these cohorts to the data sets held within NHS Digital.
The general limitations of the approach were recognised throughout. These include:
- the use of centrally held administrative data to identify patients for an intervention that could cause harm.
- the inaccuracy of the underlying data
- the incompleteness of the underlying data
- the speed at which the list was required (within 48 hours)
It was agreed that these limitations would be mitigated by local approaches, with clinical services able to add to the list locally.
In addition to the general limitations noted above, it was identified that a number of specific categories on the” High risk” list, could not be identified or fully identified by centrally held data. These are signified by a star (*) in the table above.
All code sets used for mapping clinical concepts, were derived from established clinically validated lists specifically the seasonal influenza code list (Annex G).
This list is maintained in the SNOMED CT terminology, codes identified as relating to the “high risk” categories were mapped using validated tools, and manually rechecked.
Additional conditions identified specifically by the CMO’s office were then added to the final ICD code list (note these are absent form the SNOMED CT code list contained in Annex B)
- The flu at risk list was reviewed and mapped to the CMO categories by three clinicians.
- The resulting subset was mapped via validated tooling to ICD10 by a clinical terminologist using validated tooling.
- The ICD 10 Code set was then manually re checked by two classification and terminology specialists and a clinician.
- Additional conditions as specified by the CMO’s office were then added to this code set to produce the final list.