1. Data Quality Webinar for General Practice
Ian Spiers, Terminologist, NHS Digital

2. Data Quality
Will be recording this session and if the recording is of a good enough quality we will publish the recording. I have Helen on the call as well who will help facilitate the webinar and questions.

3. Tools to be used
Webinar is for you to have a discussion about data quality issues and approaches
We will reference Data Quality Guidance
Use Look-up for Read to SNOMED
SNOMED CT Browser
Today : Use 8 points from Data Quality Guidance to discuss
Further Webinars will be based on feedback from Data Quality Leads on the Delen Forum

4. Discussion Points 1. Inactive Content 2. Terms ending in NOS / NEC
3. Abbreviations
4. Duplicate Terms
5. Outdated Terms
6. [ ] Terms
7. Terms that should be children but are not
8. Grouper Terms

5. Inactive Content
Some Read codes are mapped to inactive SNOMED CT Acute bronchitis/bronchiolitis inactive in SNOMED CT - It is ambiguous
Need to choose

6. Inactive Content
Serum folate borderline is inactive in SNOMED CT It is ambiguous, do have Borderline low and Borderline high
Need to choose

7. Inactive Content
Terms with Clinic A / Clinic B have codes mapped to inactive SNOMED CT concepts These will map but to inactive codes; they are not interoperable, never went through GP2GP Different Practices will have different clinics Use specific clinic type administrative terms in future e.g. seen in asthma clinic, referral to asthma clinic

8. Inactive Content
Some Read codes are mapped to inactive SNOMED CT Descriptions
Need to choose

9. Terms ending in NOS / NEC
Read terms ending in NOS (not otherwise specified)
This also shows that Read had synonyms that not exactly true
Otitis media is a type of middle ear disorder in SNOMED CT and not always infective
Read terms ending in NOS (not otherwise specified), NEC (not elsewhere classified), NOC
(not otherwise classified) or HFQ (however further qualified) and those beginning Other
specified originated from the classifications (ICD and OPCS). The meanings intended by
these pre- and postfixes are very specific: the meaning is also unique to one particular
version of the classifications. ICD and OPCS are typically static for a number of years. These
codes therefore serve as generic ‘catch alls’ for conditions that lack a dedicated code in the
specific version of the classification. If a dedicated code for the condition is added to a later
version of the same classification, then the meaning of these catch-all terms changes.
SNOMED CT changes every 6 months, and is used for recording details at the point of care.
As such these codes have no place in a patient record.
When Read was first designed it took content from the classifications and so these catch-all
terms also exist in Read; they have been used to code significant volumes of real patient
data. The mapping tables therefore provide a map for such codes into SNOMED CT, with
most mapping to a SNOMED synonym with the same text but without the NOS or NEC etc
or, in the case of some typically very general codes, to text which is different but still reflects
the same or very similar meaning

10. Terms ending in NOS / NEC
Read terms ending in NEC (not elsewhere classified)
Some map to exact term minus NEC – Vagotomy NEC
Some map to slightly different text minus NEC but still means the same – Revision cerebrovent shunt
Some map to a more “generic” term – Extracranial vagotomy NEC maps to Vagotomy
Which has a synonym of Extracranial vagotomy operation

11. Abbreviations
Because of the term length restrictions inherent to READ v2, some of its terms were
abbreviated. By contrast, the aim to avoid ambiguity in SNOMED CT means that such
Descriptions do not exist in SNOMED CT. The requirement on all system suppliers is to
accommodate these longer descriptions in their system design. These abbreviated terms are
mapped to an unabbreviated equivalent
Most common clinical abbreviations do exist in SNOMED CT, and can be used to search for
terms. They generally are a synonym and the full expansion of the abbreviation will normally
also appear after the abbreviation, within the same Description –
This is because some abbreviations are used in different specialties with different meanings with the same Abbreviation for example PID can be a pelvic inflammatory disease
OR a prolapsed intervertebral disc.

12. Duplicate Terms
There are a number of terms that are in Read twice; sometimes by mistake (we now have
sophisticated software to check for duplicates at the authoring stage) or because it could be
categorised in more than one way and so was placed in all possible appropriate locations.
This means two Read codes may correctly map to the same SNOMED CT Concept, but
consequently the SNOMED Description may have slightly different wording to the original
Read terms.

13. Duplicate Terms
There are a number of terms that are in Read twice; sometimes by mistake (we now have
sophisticated software to check for duplicates at the authoring stage) or because it could be
categorised in more than one way and so was placed in all possible appropriate locations.
This means two Read codes may correctly map to the same SNOMED CT Concept, but
consequently the SNOMED Description may have slightly different wording to the original
Read terms.

14. Outdated Terms
Diabetes mellitus is now described by whether it is type 1 or type 2 rather than the older
convention of describing as Insulin or Non-Insulin dependent. Data entry should use terms
with type 1 or type 2 for diabetes mellitus in Read v2.

15. [ ] Terms Read Codes with prefix [X], [SO], [M], [V], [Q], [D]
All originate from classification type terms added to Read
Codes beginning [X] exist as the same description without the [X]
Codes beginning [SO] exist as the same description without the [SO]
Codes beginning [M] exist as the same description without the [M]

16. Terms that should be children but are not
Ran out of space in Read – 5 character codes
C10E1 Type 1 diabetes mellitus
with ophthalmic complications
Ideally have the following as children
C10EF Type 1 diabetes mellitus
with diabetic cataract
C10E7 Type 1 diabetes mellitus
with retinopathy
C10EP Type 1 diabetes mellitus
with exudative maculopathy
Some terms currently published in Read v2 should be classified under some existing level 5
term … but they cannot be because of the Read v2 hierarchy limit. Instead they have had to
be represented used the fixed 5 character codes that Read v2 provides.

17. Grouper Terms
Within Read v2 some codes exist that serve only to gather together codes related to the
same clinical area so that they’re easier to find when using the hierarchy to pick a code for
data entry. Therefore, they often gather together codes to record both when a given
phenomenon is ‘present’ and when it is ‘not present’. This is NOT the case in SNOMED CT;
codes where the symptom or condition is absent do exist in SNOMED but in a completely
separate hierarchy . These grouper terms make it difficult to analyse data in Read as the children
are a mix of codes

18. Collaborative space - Delen
Further Webinars based on feedback from Data Quality Leads
Use Delen
CCG Nominated Leads - SNOMED CT in Primary Care
SNOMED CT in Primary Care Trainers
Workspace discussion forum for raising issues
If you don’t have access

19. Webinar Roundup
Any questions?