search

opensafely / codelist-development

Repository for discussion of OpenSAFELY codelists
7 stars 4 forks source link

*CLINICAL CONDITION* Other conditions affecting immunity #11

Open CarolineMorton opened 4 years ago

CarolineMorton commented 4 years ago

See lshtm code list:

HIV_Jul19.xlsx other_cmi_immuno_updated_Jul18.xlsx

alexwalkerepi commented 4 years ago

In light of ebmdatalab/tpp-sql-notebook#58 I've grouped a few different diseases together here:

alexwalkerepi commented 4 years ago

DRAFT

DEFINITION: Immunosuppressing conditions not covered elsewhere: aplastic anaemia, HIV, genetic conditions that increase risk of infections, and non-specific codes recording a state of immunosuppression (such as "Patient immunosuppressed")

POTENTIAL BIASES: Unclear what underlying conditions the 'immunosuppression NOS' records may capture, and duration of immunosuppression following them - any time period cut-off (or none) likely to misclassify some patients' immunosuppression status based on these codes.

CLINICAL SIGN OFF & DATE:

EPIDEMIOLOGY SIGN OFF & DATE:

SHARED WITH WIDER TEAM: Yes/No

FINAL SIGN OFF DATE (and apply label)

hmcd commented 4 years ago

@CarolineMorton @alexwalkercebm have finished the QOF search and updated Alex's definition above as an edit - SNOMED to follow

hmcd commented 4 years ago

@CarolineMorton aplastic QOF codes - hopefully it works this time! aplastic_QOF.xlsx

alexwalkerepi commented 4 years ago

CTV3 codlists back from TPP: Immuno_Aplastic_CTV3_Raw.xlsx Immuno_HIV_CTV3_Raw.xlsx Immuno_Genetic_CTV3_Raw.xlsx Immuno_NOS_CTV3_Raw.xlsx

amirmehrkar commented 4 years ago

Immuno_Aplastic_CTV3_Raw - Amir.xlsx

@CarolineMorton @alexwalkercebm have identified some issues - it's mainly to do with the definition of aplastic anaemia being all 3 cell lineages in the bone marrow being affected, and this codelist also has several codes about "rec cell" lineage only/anaemia alone.

I dont think they are similar, esp as red cell lineage wouldn't confer immunosuppression like aplastic. So I would remove those. Happy to discuss. Ideally, best to confer with a haematologist. If there is an anaemia codelist, I would put these there.

alexwalkerepi commented 4 years ago

Immuno_HIV_CTV3_Reviewed.xlsx

HIV relatively straightforward, just a few codes flagged for removal.

alexwalkerepi commented 4 years ago

Immuno_Genetic_CTV3_Reviewed.xlsx

Most of the exclusions here are conditions that could also be acquired as well as sometimes genetic.

alexwalkerepi commented 4 years ago

Immunosuppression NOS immune_nos_other_causes_removed.xlsx

The snomed codes I did for this I kept deliberately broad, then I wrote a quick script to remove the codes from the above 3 codelists. If we change any of the above lists, I'll rerun it, then it needs a sense check.

HenryDrysdale commented 4 years ago

Hi all, @alexwalkercebm, I've reviewed the 4 code lists below. With the exception of genetics, where I haven't commented that means I agree:

  1. Aplastic anaemia - a few comments for review, almost always agree with Amir Immuno_Aplastic_CTV3_Raw.-.Amir, Henry.xlsx

  2. HIV - mostly straightforward, just a handful of comments for review Immuno_HIV_CTV3_Reviewed, Henry.xlsx

  3. Genetic conditions - way out of my comfort zone I'm afraid! Have made a few comments, but really not sure which of these may be aquired... Immuno_Genetic_CTV3_Reviewed_Henry.xlsx

  4. Immunosuppressed NOS - I've included all, because they all imply immunosuppression. But some codes will be included in the other lists. immune_nos_other_causes_removed_Henry.xlsx

Henry

CarolineMorton commented 4 years ago

Hi @HenryDrysdale and @alexwalkercebm

I've reviewed the 4 code lists below. I agree with pretty much everything.

Aplastic anaemia: Immuno_Aplastic-AM_HD_CM.xlsx

HIV: Immuno_HIV_AM_HD_CM.xlsx

Genetic conditions: I haven't changed anything here as I lack domain knowledge but it general looks ok.

Immunosuppressed NOS: immune_nos_other_causes_removed_Henry-cm.xlsx

Caroline

hmcd commented 4 years ago

Hi @HenryDrysdale , @alexwalkercebm and @CarolineMorton

Reviewed aplastic anaemia and genetic causes of immunosuppressiong mappings.

Aplastic anaemia Agree with all the comments, suggested a further two codes likely to be red cell specific to be removed Immuno-Aplastic_AM_HD_CM_HM.xlsx

Genetic conditions I would suggest renaming this as "Permanent immunosuppression excluding HIV" and including conditions which may be acquired, if they are immunosuppressing. The purpose of this list is to be used in the immunosuppression algorithm to identify permanently immunosuppressed patients - we're not looking specifically at genetic causes as an exposure of interest . So I think we can tolerate that some conditions may have been acquired if we think the patient should still be defined as immunosuppressed- otherwise we need another codelist to include 'other permanent causes of immunosuppression'. We're using the PHE definition of immunosuppression for this algorithm, and it specifically includes complement disorders in the risk groups for flu and PPV so I would include these in a potential risk group for Covid-19 too. Immuno_Genetic_CTV3_Reviewed_Henry_CM_HM.xlsx I'm not a topic expert! Mostly worked off this https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7082301/

Not reviewed HIV, as it seems uncontroversial.

On the Immunosuppressed NOS - these only get defined as immunosuppressed for a short period of time (we use 3 months in our algorithm). So I think we can include agranulocytosis - but should we rename the variable to eg "Short term or unspecified immunosuppression" and make it clear in the definition that these are not necessarily permanent? This is quite an important list - it picks up a lot of the patients in the immunosuppressed group.

hmcd commented 4 years ago

Thought it might be helpful to summarise what these codelists aim towards - essentially trying to identify immunosuppressed patients at a single point in time, so we want separate codelists for conditions where if it is ever recorded the patient is always immunosuppressed, and conditions where we would consider the patient immunosuppressed if the code is recorded recently (e.g. 3 months though it is somewhat arbitrary). The year-24 months conditions have been put in a separate github issue (e.g. bone marrow transplant).

Based on diagnoses Codelists this corresponds to
Any history of HIV, asplenia or dysfunction of the spleen (including sickle cell disease), or other permanent immunosuppression (such as genetic conditions compromising immune function) HIV, dysplenia sickle cell, genetic conditions EVER
Any history in the previous year of: aplastic anaemia, leukaemia, lymphoma, receiving a bone marrow transplant, or receiving chemotherapy or radiotherapy. Aplastic anaemia, haematological malignancies, bone marrow transplant, chemotherapy/radiotherapy   IN LAST YEAR
Any record within the previous 3 months of a status of immunosuppression without further details. Immunosuppression NOS IN LAST 3 MONTHS
hmcd commented 4 years ago

@CarolineMorton @HenryDrysdale @alexwalkercebm looking at the Immunosuppression NOS - this includes quite a few permanent conditions. Would you be happy if I move them to 'genetic' and then we rename the variable as 'other permanent immunosuppressing conditions including genetic' - or shall I make a new list?

CarolineMorton commented 4 years ago

Yes that makes sense. I would be very happy for you to do this. With the table above, we should probably move agranulocytosis back into as I am happy for this to be included if this is in the last 3 months. Are you able to do this @hmcd, otherwise i can do this tomorrow?

CarolineMorton commented 4 years ago

FINAL SIGN OFF

DEFINITION: Immunosuppressing conditions not covered elsewhere: aplastic anaemia, HIV, permanent conditions that increase risk of infections, and temporary immunosuppression.

CODE LISTS: code lists available in V2 --> all need mapping to V3.

Originally we had planned to separate genetic conditions that increase risk of infections and immunosuppression NOS, and therefore did our searches according to this:

We then ended up with two lists, which were reviewed as above. On further discussion it was agreed that they should be split into permanent and temporary immunosuppresssion. The lists were combined and then @hmcd and @LiamSmeeth reviewed. This is the final reviewed list: Other_immunosuppression_dedup_210420 Liam.xlsx. @LiamSmeeth suggested a timeframe of 12 months for immunosuppression after one of the temporary codes.

@CarolineMorton then sorted theses 2 lists. There was 2 remaining queries which were 1) 'transplant immunosuppression' --> doubled checked and not in solid organ or bone marrow transplantation so kept in this list. 2) [M] T-gamma lymphoproliferative disease - not in haematological cancer as not malignant so was kept in permanent list .

FLEXIBILITY NEEDED BETWEEN STUDIES: May want to apply varying time periods to define immunosuppression based on aplastic anaemia (suggested standard 24 months) and the 'immunosuppression NOS' category (suggested standard 3-12 months depending on need for sensitivity/specificity).

EFFECTS ON COHORT SELECTION: None

POTENTIAL BIASES: Unclear what underlying conditions the 'immunosuppression NOS' records may capture, and duration of immunosuppression following them - any time period cut-off (or none) likely to misclassify some patients' immunosuppression status based on these codes.

CLINICAL SIGN OFF & DATE: Liam Smeeth 24th April 2020 08.15

EPIDEMIOLOGY SIGN OFF & DATE: Helen McDonald 23 April 2020 20:47

SHARED WITH WIDER TEAM: Yes/No

FINAL SIGN OFF DATE (and apply label)

LiamSmeeth commented 4 years ago

Caroline, Helen Yes this all looks correct and sensible to me – I like the flexibility for future studies. Sorry to be dim, but do I need to do anything to help get these signed off? Thanks L

From: CarolineMorton notifications@github.com Reply to: ebmdatalab/tpp-sql-notebook reply@reply.github.com Date: Thursday, 23 April 2020 at 20:08 To: ebmdatalab/tpp-sql-notebook tpp-sql-notebook@noreply.github.com Cc: Liam Smeeth Liam.Smeeth@lshtm.ac.uk, Mention mention@noreply.github.com Subject: Re: [ebmdatalab/tpp-sql-notebook] CLINICAL CONDITION Other conditions affecting immunity (#36)

FINAL SIGN OFF

DEFINITION: Immunosuppressing conditions not covered elsewhere: aplastic anaemia, HIV, permanent conditions that increase risk of infections, and temporary immunosuppression.

CODE LISTS: code lists available in V2 --> all need mapping to V3.

Originally we had planned to separate genetic conditions that increase risk of infections and immunosuppression NOS, and therefore did our searches according to this:

We then ended up with two lists, which were reviewed as above. On further discussion it was agreed that they should be split into permanent and temporary immunosuppresssion. The lists were combined and then @hmcdhttps://github.com/hmcd and @LiamSmeethhttps://github.com/LiamSmeeth reviewed. This is the final reviewed list: Other_immunosuppression_dedup_210420 Liam.xlsxhttps://github.com/ebmdatalab/tpp-sql-notebook/files/4524661/Other_immunosuppression_dedup_210420.Liam.xlsx.

@CarolineMortonhttps://github.com/CarolineMorton then sorted theses 2 lists. There was 2 remaining queries which were 1) 'transplant immunosuppression' --> doubled checked and not in solid organ or bone marrow transplantation so kept in this list. 2) [M] T-gamma lymphoproliferative disease - not in haematological cancer as not malignant so was kept in permanent list .

FLEXIBILITY NEEDED BETWEEN STUDIES: May want to apply varying time periods to define immunosuppression based on aplastic anaemia (suggested standard 24 months) and the 'immunosuppression NOS' category (suggested standard 3 months).

EFFECTS ON COHORT SELECTION: None

POTENTIAL BIASES: Unclear what underlying conditions the 'immunosuppression NOS' records may capture, and duration of immunosuppression following them - any time period cut-off (or none) likely to misclassify some patients' immunosuppression status based on these codes.

CLINICAL SIGN OFF & DATE:

EPIDEMIOLOGY SIGN OFF & DATE:

SHARED WITH WIDER TEAM: Yes/No

FINAL SIGN OFF DATE (and apply label)

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/ebmdatalab/tpp-sql-notebook/issues/36#issuecomment-618599877, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APDUVH7K4I7IFONMSVVYT2LROCGZ5ANCNFSM4LXTSLJQ.

CarolineMorton commented 4 years ago

just edit the box above: https://github.com/ebmdatalab/tpp-sql-notebook/issues/36#issuecomment-618599877 (direct link here)

click the 3 little dots on the top right corner and click edit. you should then be able to edit the text. if you could add your name to clinical sign off and the date, that would be fantastic. thank you @LiamSmeeth

LiamSmeeth commented 4 years ago

Done I think!

From: CarolineMorton notifications@github.com Reply to: ebmdatalab/tpp-sql-notebook reply@reply.github.com Date: Friday, 24 April 2020 at 08:13 To: ebmdatalab/tpp-sql-notebook tpp-sql-notebook@noreply.github.com Cc: Liam Smeeth Liam.Smeeth@lshtm.ac.uk, Mention mention@noreply.github.com Subject: Re: [ebmdatalab/tpp-sql-notebook] CLINICAL CONDITION Other conditions affecting immunity (#36)

just edit the box above: ebmdatalab/tpp-sql-notebook#36 (comment)https://github.com/ebmdatalab/tpp-sql-notebook/issues/36#issuecomment-618599877 (direct link here)

click the 3 little dots on the top right corner and click edit. you should then be able to edit the text. if you could add your name to clinical sign off and the date, that would be fantastic. thank you @LiamSmeethhttps://github.com/LiamSmeeth

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/ebmdatalab/tpp-sql-notebook/issues/36#issuecomment-618845351, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APDUVH2CNDZQDGUE3ITBOKLROE3ZRANCNFSM4LXTSLJQ.

krishnanbhaskaran commented 4 years ago

The codes XaFuN, XaFuO and 43CZ. appeared in early analyses to represent HIV testing rather than HIV positivity, including widespread use in young women, likely at pregnancy. With these codes, numbers with HIV were far higher than expected given national data. After discussions the code list has been edited, removing these codes.