*MEDICINE*: Oral steroids

[sebbacon]

Distinguish between long term and acute treatment

[CarolineMorton]

Code list from LSHTM:

oralsteroids_drugcodes_Jul19.xlsx

[CarolineMorton]

@brianmackenna Are you able to sense check these existing code lists? Do you think there is any point running our own with what we already have from OP?

[brianmackenna]

Here's a steroid notebook I prepared earlier! These aren't directly comparabale - finalising list will depend on the criteria we hammer out.

Quick skim but keen for @inglesp view - I think our SnoMed/dm+d generation process might be more efficient. Updates are applied weekly and we have already automated links to prescribing data plus all the fields in dm+d.

[inglesp]

@brianmackenna I don't have enough context to understand what might be more efficient. Chat on the phone?

[brianmackenna]

I have reviewed the xlsx file above and I propose we generate code lists using our dm+d databases and where an previously curated codelist exists we use this for sense/checking validation. Our dm+d databases gve us advantage of a) it allows a huge degree of automation b) I think the dm+d is more comprehensive and is used in TPP (although it sits within Multilex) c) Our database updates on a weekly basis d) it allows us to use all fields in the dm+d to make our list more specific e.g. All preparations starting with BNF code 0501 that are administered is a solutioninjection

[sebbacon]

(have added new issue ebmdatalab/tpp-sql-notebook#48 for capturing consensus around using dynamic medicine codelists)

[brianmackenna]

Steroids specifically discussed on a call today with DataLab @CarolineMorton @amirmehrkar @alexwalkercebm etc and @hmcd

We can align the "replacement steroids" (note they are also used for acute flares/crisis) and xlsx file above.

We then can apply logic to see if we can distinguish between

• short courses/rescue packs/flares
• low dose long term
• rarer disease and where used in combination with DMARDs etc.

Using dose syntax is likely to be long and fiddly so pragmatically we will explore using

• n number of issues in a time period
• quantity issued in a time period

[alexwalkerepi]

Draft definition

DEFINITION:

CODE LISTS: FLEXIBILITY NEEDED BETWEEN STUDIES: EFFECTS ON COHORT SELECTION:

POTENTIAL BIASES:

CLINICAL SIGN OFF & DATE:

EPIDEMIOLOGY SIGN OFF & DATE:

SHARED WITH WIDER TEAM: Yes/No

FINAL SIGN OFF DATE (and apply label)