Question: TRT01P in ADSL

[ynsec37]

Dear team,

The current ADSL example derives the No Treatment for TRT01P that do not have a treatment

adsl <- adsl_raw %>%
  mutate(
    TRT01P = if_else(ARM %in% c("Screen Failure", "Not Assigned", "Not Treated"), "No Treatment", ARM),
    TRT01A = if_else(ACTARM %in% c("Screen Failure", "Not Assigned", "Not Treated"), "No Treatment", ACTARM)
  )

In the below ADaM IG, The TRT01P is likely not derived for "Screen Failure" at least
https://www.cdisc.org/system/files/members/standard/foundational/[ADaMIG_v1.3.pdf]

Although I have seen both scenarios that some derive it and some not, I just wonder there is a general rule.
Could you kindly help to provide your option? Thank you very much.

• if TRT01P should be derived for 'Screen Failure'

• if TRT01P should be derived for Randomized but "Not Assigned", "Not Treated"

[bms63]

I usually see TRT01P with Not Treated. I think there is a new DM.ARMRS variable that captures "Screen Failure" and this is no longer present in DM.ARM but a bit fuzzy on it. But we don't have that in our DM data...maybe worth do modernize depending on other responses.

@nbrucken17, @manciniedoardo or @bundfussr WDYT?

[bundfussr]

I think it makes sense to modernize our DM data and simplify the derivation in the ADSL example.

[bms63]

@Fanny-Gautier and @Lina2689 what do you all think? update for pharmaversesdtm?

@rammprasad would any examples in sdmt.oak be impacted - im thinking no, but maybe you could weigh in here as well.

[Fanny-Gautier]

@Fanny-Gautier and @Lina2689 what do you all think? update for pharmaversesdtm?

@rammprasad would any examples in sdmt.oak be impacted - im thinking no, but maybe you could weigh in here as well.

Both DM.ARMNRS and DM.ACTARMUD could be added in {pharmaversesdtm}

[bms63]

Shall I make an issue in pharmaversesdtm?

Is anything in pharmaverseraw needing to be updated just to keep consistency? @rammprasad - or @ShiyuC any ideas?

[nbrucken17]

Hi all,

As @bms63 mentioned, the variable ARMNRS (Reason Arm and/or Actual Arm is Null) was introduced in SDTMIG v3.3 as a place to store information on why ARM and/or ACTARM are null for a given subject. It's generally populated with values such as "SCREEN FAILURE", "NOT ASSIGNED", "ASSIGNED, NOT TREATED", "UNPLANNED TREATMENT", etc., and either ARM, ACTARM or both would be missing in such cases. That's a change from earlier versions of SDTM where "Screen Failure" was stuffed into ACTARM. If the DM information does not map screen failures, etc. according to current SDTMIG standards, I think it should be updated.

As for ADaM, unless screen failures are summarized in a separate column on the TFLs, I would not populate either TRT01P or TRT01A with "Screen Failure". "Screen Failure" and "Not Assigned" are definitely not planned treatments (nobody enrolls a subject into a study planning for them to be a screen failure), and unless summarized as such, are not actual treatments either.

For subjects who are randomized but not treated, TRT01P is generally populated with the treatment to which they were assigned, while TRT01A would be left missing, since they didn't actually receive anything. TFLs displaying actual treatment groups are often run against the safety population, defined as subjects who actually received at least 1 dose of study medication, so this approach works because subjects with missing TRT01A are excluded from those TFLs anyway.

Hope this helps,
Nancy

[Fanny-Gautier]

Shall I make an issue in pharmaversesdtm?

Is anything in pharmaverseraw needing to be updated just to keep consistency? @rammprasad - or @ShiyuC any ideas?

dm domains are usually created from below files

# Create dm, suppdm ----
sdtm_path <- "https://github.com/cdisc-org/sdtm-adam-pilot-project/blob/master/updated-pilot-submission-package/900172/m5/datasets/cdiscpilot01/tabulations/sdtm/" # nolint
raw_dm <- read_xpt(paste0(sdtm_path, "dm", ".xpt?raw=true"))
raw_suppdm <- read_xpt(paste0(sdtm_path, "suppdm", ".xpt?raw=true"))

I guess @Lina2689 (or someone else) will need to add manually DM.ARMNRS and DM.ACTARMUD.

If you are fine with the approach you can create the corresponding issue in {pharmaversesdtm} . Thank you.

[bms63]

Thanks all for great discussion. I made an issue in pharmaversesdtm.

Let's leave this one open and will make update to examples site once completed. @Lina2689 and @Fanny-Gautier this isn't for the current pharmaveresdtm release!! for the next release!!

[bms63]

BTW - @ynsec37 how do we get you more involved with pharmaverse!?! You bring such great issues love to get you on any team!!

[ynsec37]

Thank you all for your time and valuable discussion.

Hi @bms63, you can find me in the pharmaverse slack. I am happy to be involved.