Update derive_agegr_fda categories: <18, 18-64, >=65

[millerg23]

derive_agegr_fda has link to https://prsinfo.clinicaltrials.gov/results_definitions.html that states age categories are <=18, >18 and <65, >=65. Both Roche and GSK use categories <18, 18-64, >=65. The link provided is not relevant, and should be removed and use categories <18, 18-64, >=65.

[bms63]

@tomratford This might be a good one for you to tackle next. Gets you more into the code and rebuilding the roxygen documentation.

[thomas-neitmann]

@tomratford We decided to have issues automatically closed once the changes make it to the main branch (rather than just devel).

[aehmann-gsk]

@thomas-neitmann @millerg23 @bms63 @bundfussr @tomratford This function is not correct for GSK. When results are disclosed, they must align with the guidance provided here: https://prsinfo.clinicaltrials.gov/results_definitions.html The link provided is relevant and correct.

(Note: you can see trial results for the FDA disclosed here by searching for studies, it is kind of cool - you can search on a study and if the results are disclosed you can see an overview of the study, results must be disclosed within a period after an analysis (the window is escaping me but I think it is 6 months)): https://clinicaltrials.gov/ct2/results?cond=&term=a&cntry=&state=&city=&dist=&Search=Search

Age must be disclosed per clinicaltrial.gov in 1 of the 3 ways: Age * (Select at least one of the following): Age, Continuous: For example - mean or median age Age, Categorical: <=18 years

18 and <65 years =65 years Age, Customized: Customizable age categories

I am not sure why the default FDAAA groupings were changed but GSK computes these groupings as <=18, 19-64, >=65.

Is there a reason this was changed? It does not align with GSK standards nor ClinicalTrials.gov which is what this grouping is used for.

[bms63]

@rossfarrugia Do you have any thoughts or remember on how we arrived at this decision? I think Gordon is out of town, but it looks like he created the original issue.

[bundfussr]

The topic is on the agenda of the next Questions and Comments meeting. See also https://github.com/pharmaverse/admiral/issues/856.

[rossfarrugia]

I was talking to @millerg23 about this the other week as we found Roche actually uses "<18", "18 - 65", ">65", so everyone seems to be in some way doing it slightly differently. Our consensus was just maybe better to deprecate the FDA and EMA versions in admiral, and just give people a more generic agegr function where they can easily provide their own required groupings. Either would be using case_when as in our current ADSL vignette examples or get_map as per whatever gets agreed from this discussion: https://github.com/pharmaverse/admiral/issues/856

[millerg23]

yes agree with @rossfarrugia . There doesn't seem to be a standard age grouping defined by FDA. If we want to keep the original function, maybe it needs renamed to something else, something like derive_agegrp_ctgov and include the original link? A generic age group function is required also.

[rossfarrugia]

To close this out: see https://github.com/pharmaverse/admiral/issues/856 where we will now deprecate these and instead of having a generic age group function we'll use case_when as in ADSL vignette example currently.