Validation of CMDECOD values >200 characters

n Nancy

on December 10, 2024

Long CMDECOD text is correctly being split into CMDECOD, CMDECOD1, etc., as specified in the SDTMIG. The problem is that these split CMDECOD values are being flagged as not appearing in the WHODrug dictionary. From this end, it appears that P21 is only checking CMDECOD itself against the contents of the WHODrug dictionary, so looking at just the first 200 characters, without concatenating the values found in the additional segments. Is that the case, and if so, is there a workaround for it besides manually confirming the full term appears in the dictionary, and then documenting the false positive in the cSDRG?

Thanks,
Nancy

Forums: SDTM

j Jozef

on December 11, 2024

Validation of CMDECOD values >200 characters

Dear Nancy,

I first was a bit confused about your statement about CMDECOD1, so I did some research.
What currently (unfortunately) needs to be done is that the first 200 characters (or less, until the last semicolon in the "list") go into the CM dataset, variable CMDECOD.
What is remaining, must then go in chunks of 200 characters (or less, as the splits need to be done on semicolons, rather than words), go into SUPPCM, with QNAM = CMDECOD1, CMDECOD2, etc..

For validation software, this means that before testing whether the values are appearing in the WHODrug dictionary, the QVAL values in SUPPCM for QNAM = CMDECOD1, CMDECOD2 etc, must be "reunited" with the value of CMDECOD in CM. This is a requirement that essentially applies to all values for any variable that is longer than 200 characters.
You say that P21 validator seems not to do so, so I would (when this is so) consider this as a software bug. I will do some tests.

Exactly such issues are the reason why this industry and regulatory authorities must get rid of SAS-XPT as soon as possible. 10 years ago a better format, Dataset-XML was proposed to the FDA, but it never got it into the "Standards Catalog" out of fear of some FDA people that file sizes could considerably get larger (it seems submissions are not immediately loaded into databases at FDA as it should be, but are even kept as "files" on disc or memory stick). A new effort of CDISC has now resulted in a new format, Dataset-JSON that has none of the issues of SAS-XPT, e.g. no "200-character limitation" and has small file sizes (but FDA should load submissions into databases, and not keep them on file anyway). So we hope that FDA and other regulatory authorities will soon start accepting Dataset-JSON. This of course also means that validation tools will need to be adapted, but that is exactly what is currently happening in the CORE (CDISC Open Rules Engine) project by CDISC.
So there is hope ...