Validator 1.4 new validation criteria?

Pinnacle 21 Support SEND
1

Hi Amelia,

Thanks for identifying a standard metadata definition issue for SEDN validation. We don’t use external CSV files as a source for standard metadata anymore. Unfortunately for SEND this transition was not very successful for several reasons. The major one is a lack of SEND data for testing purposes, which lead us to miss this config specs bug during testing.

We are working to fix SEND metadata and replace existing package ASAP.

  1. It’s a bug with incorrect SEND metadata.

  2. According to SDTM IG and SEND IG the only allowed values for variables of SD1054 check are ‘Y’ and NULL. 'NA" value should not be used there.

  3. See “CDER Common Data Issues” document. It is considered as a FDA Guideline.

http://www.fda.gov/downloads/Drugs/DevelopmentApprovalProcess/FormsSubmissionRequirements/ElectronicSubmissions/UCM254113.pdf

See page 12

  1. Yes. it’s quite controversial. However we decided to keep it like this. This issue is still under discussion. I think, that --TEST, --TESTCD, etc. variables should be excluded from SD1080-82 checks and still have their own exclusive business rules.

  2. It’s a metadata bug in our specs

  3. It’s a metadata bug in our specs

  4. I am not sure. It’s a part of “CDER Common Data Issues" document and PhUSE working group request. The actual business rule is that you need to have both Study Reference Start Date and Data Collection Date values, which both include complete date part to calculate Study Day. Otherwise it’s considered as imputation of data values, which is not allowed in SDTM/SEND.

Regards,

Sergiy

2

Hi Sergiy,

Thank you for the clarifications.

I’m looking forward to test the updated package.Please let me know when it’s available and I would be more than happy to give it some tests and to report anything that looks “suspicious” to me :slight_smile:

Thanks again. Keep up the good work!

Amelia

3

Thank Amelia!

The real reason we got this issue with SEND profile is that we do not have enough real SEND data because SEND adoption level is very low compare to SDTM. As result out test cases did not catch some wrong SEND metadata specifications.

We will be very happy to provide updated version of SEND validation profile to you before official update. I’ll contact you later.

Thank you again!

Sergiy

4

Hi Sergyi,

Some observations based on the latest downloadable version. I see some changes in the SEND section, but I’m still noticing some issues:

  • BG: BGSTRESC variable not added in BG ItemGroupDef => SE0058 Variable appears in dataset, but is not in SEND standard

  • DS: DSDECOD still not added in ItemGroupDef for DS => SE0058 Variable appears in dataset, but is not in SEND standard

  • BG: VISITDY added but not in SEND standard => SE0057 SEND Expected variable not found

  • TS: STUDYID and DOMAIN are in wrong order => SD1079 Variable is in wrong order within domain

  • MI: MILAT label should be “Specimen Laterality within Subject”

  • LB: LBORNRLO, LBORNRHI should have numeric type, not text in ItemDef entry.

  • LB: LBDY, LBENDY have different labels than the ones from SEND specs.

  • SE2201 should have same Severity as SE2202, SE2203, etc (“Warning”).

  • SE0015, SD0026, SD0029 should be disabled for SC and PM - see older discussion: http://www.opencdisc.org/forum/se0015-missing-stresn-value-and-sd0026-missing-units-value

  • TA: some ItemRef are not in order. When define.xml not specified => SD1079 - Variable is in wrong order within domain.

  • Error SD0037 Value for TXPARMCD not found in (SEND Trial Summary Parameter Test Code) user-defined codelist is wrongly reported for TXPARMCD = ‘GRPLBL’.

Best regards,

Amelia

5

Disregrad observation: “- LB: LBORNRLO, LBORNRHI should have numeric type, not text in ItemDef entry.”

The type is numeric in SEND 3.0 but apparently it will change to text in 3.1

6

Hi guys,

First of all, congratulation for the new release!
I just downloaded the 1.4 version and re-validated some datasets that I tested before using the 1.3 version. I’m getting some new Errors/Warnings - maybe someone can clarify, please?

#1. SE0063 - “SEND/dataset variable label mismatch” for some BW, SC, DM, CL, MA, MI, TF, TX variables (BWTESTCD, BWTEST, BWORRES, BWEXCLFL, SCGRPID, TXPARMCD, etc).
By looking in config-send-3.0.xml I see the labels there are different from the ones in my dataset, which match the labels from SENDIGv3.pdf. Are the rules are defined against a new SEND version? If so - which one?

#2. SD1054 - “Non-recommended variable length” - “The length of flags variables (----FL, --FAST, --OCCUR, --PRESP) is expected to be 1”.
I thought it was decided some time ago to change the length from 1 to 2 to accomodate the “NA” value?

#3. SD1071 - “Dataset is greater than 1 GB in size” - “Large datasets should be splitted into smaller datasets no larger than 1 GB in size.”
Why this limitation?

#4. SD1080 / SD1081 / SD1082 - “-- variable length is too long for actual data” - “Variable length should be assigned based on actual stored data to avoid to minimize file size. Datasets should be resized to the maximum length used prior to splitting.”
I’m not sure I understand the message - does this rule cancel the rules like --TESTCD, ETCD, QNAM, etc is always on 8 characters, --TEST, QLABEL etc is always on 40 characters, etc?

#5. SE0057 - “SEND Expected variable not found” for BG: VISITDY. Same question as at point #1: the rules are defined against a new SEND version?

#6. SD0059 - “Define.xml/dataset variable type mismatch” for some variables like
a) BGSTRESC - that is not even defined in config-send-3.0.xml (hence also another warning: “SE0058 - Variable appears in dataset, but is not in SEND standard”);
b) DSDECOD - which again is not defined in config-send-3.0.xml but referenced in rule “CT1039”.

#7. SD1085 - “–DY variable value is imputed” for CODY. Should this rule really apply to DY from Comments?

Best regards,

Amelia