*CONDITION*: HIV

[a-d-brown]

User Requirements The ISARIC COVID-19 case report form collects data on patients medical history. A detailed completion guide collects data on patients medical history. A planned ISARIC validation study requires a series of SNOMED CT codelists to replicate as closely as possible the details given in the ISARIC case report form. This codelist will be used to extract patients with HIV as defined by ISARIC criteria.

Search terms / strategy All appropriate codes already identified in NHSD HIV_COD refset.

Inclusion Criteria for Codes

• Listed in the NHSD HIV_COD refset.

Exclusion Criteria for Codes

• Not listed in the NHSD HIV_COD refset.

Link to Codelist

• https://www.opencodelists.org/codelist/user/Andrew/hiv/1efade93/

[speed-vm]

@a-d-brown Think we need to add codes with Acquired immune deficiency syndrome 62479008 (160 uses in 2022). Instead of saying identify those with HIV in the method, suggest to rephrase due to sensitivity, perhaps extract?

[a-d-brown]

Thanks @speed-vm, I've made those changes. Let me know if that looks ok, then I'll publish.

[speed-vm]

@a-d-brown Yes looks good to me

[brianmackenna]

Can I ask why we didn't use the OpenSAFELY codelists that were used in the HIV study we did?

https://www.opencodelists.org/?q=hiv

[a-d-brown]

Me and Chris discussed early on that we would aim to use an NHSD Refset where one existed. This was partly to try and give consistency to the approach, and a definitive source which can be referenced in the study.

It's also because, in general we found that the NHSD Refsets were the most comprehensive lists available. We noticed some quirks in some other codelists that have been automatically converted to SNOMED CT from CTV3 (such as the one you've mentioned); often many specific but relevant codes seem to be excluded within the tree. We assumed this to be an artifact of the conversion process. These omissions don't seem to occur with the NHSD Refsets.

[brianmackenna]

I think we need to capture some of this and discuss with data team at some point! It sounds like there might be important learning