Rule AD0078 - AD0079

RULE AD0078 - AD0079: answer

Thanks Natalie.    According to the IG page 15, only TRT01P is required and TRxxSDT and TRxxEDT are CONDITIONALLY REQUIRED.  These validations target the CONDITIONALLY REQUIRED aspect and only make sure that TRxxSDT and TRxxEDT exist (don't have to be populated) if TRTxxP is present.   Since TRT01P is required, then at least TR01SDT and TR01EDT will be required as well.   

In the short term, if you want them to be treated as Warnings, you can change the Type in the config file.  However the ADaM Team has reported that all validations be treated as Errors.  I will forward this to the ADaM team.   thanks again.

Mike DiGiantomasso, Pinnacle 21

Rule AD0078-AD0079: answer

HI Mike,

I would disagree at this point. Surely if you check the example on page 40, you are right.

However, both variables TRTSDT and TRTEDT are also conditional. I would say, that if only TRT01P is populated it would be sufficient to present TRTSDT and TRTEDT. Otherwise we would present 4 variables with the same information. If more treatment variables are available, means TRT01P and  TRT02P are presented, I would expect TRxxSDT and TRxxEDT.  I think the guideline can be interpreted this way. What do you think.

Regards, Natalie

ADSL Treatment Dates

  Are you referrring to the 3rd example (Table 4.1.3)

1. treatment variables for 3 subjects in a three-period cross-over design.
2. TRTSDT , TRTEDT: not displayed
3. TRTSDT= TR01SDT
4. TRTEDT= MAX(TR01EDT, TR02EDT, TR03EDT) as some subjects may discontinue before TRT03P.

I do agree that if only TRT01P is populated, TRTSDT and TRTEDT are enough.   TR01SDT and TR01EDT would be duplicates.   The ADaM team tried to make the rules simple, generic and machine readable by all programming languages.   This may be the reason more complex logic wasn't built in to some of the rules.

However since TRTSDT and TRTEDT are conditional and the rules ensure that TRT01P, TR01SDT, TR01EDT are required...you could leave TRTSDT and TRTEDT null in most cases.

these would all be valid

thanks

Mike D

ADSL treatments dates

HI Mike,

I'm not sure why the TRxxSDT and TRxxEDT are required.

According to the guideline:

The availability of the variable TRT01P doesn't mean that there are multiple treatments nor that multiple preiods exist. In the case that only TRT01P is available the TRxxSDT and TRxxEDT are conditional too.

Of course it is possible to leave the variables TRTSDT and TRTEDT null as you suggest, on the other side, ADSL usualy has more than 50 variables and I would like to avoid to add any variables which are not needed or has values null.

In my previous replay I was referencing to the examlpe 4.1.1.

Kind regards,

Natalie

That's true.  I agree that

That's true.  I agree that the prescence of TRT01P does not mean there are multiple treatment periods.   I think this issue boils down to what one considers the authoritative source of rules.  

According to the CDISC ADaM Validation Checks Version 1.0 :  the *Machine-Testable Failure Criteria for rules 78 and 79 are as follows:

• TRTxxP is present and TRxxSDT is not present
• TRTxxP is present and TRxxEDT is not present

These statements are cross-referenced with a field called "Text from ADaM IG" which is defined as the requirement paraphrased from the ADaM IG.  in this case, both 78 and 79 are associated to this statment:

• If there is more than one treatment period then TRxxSDT and TRxxEDT should exist in ADSL.

When going through the various rules, i used the Machine-testable failure criteria as the main source for implementation, and then the IG to supplement.   there was no easy way to only implement 78 and 79 when multiple TRTxxP variables existed.   I interpreted that the ADaM team felt the same way which is why they did not qualify the machine-testable failure criteria.   

I'll certainly note this good feedback and take it back to the ADaM team.   For now, the rule simply errs on the side of caution.   Requiring consistent variable pairings in cases where it might be redundant is safer (maybe not better) than allowing them to be missing when they are really needed.

* Machine-Testable Failure Criteria (Rule)

a programmable test, written such that an affirmative response represents a failure of the requirement. This text is intended for use as a requirement specification which could be implemented in a variety of programming languages.

Rule AD0078-AD0079

Thar's fine.

Thanks,

Natalie

You're welcome.   We

You're welcome.   We appreciate all feedback (especially with less mature standards such as ADaM and SEND) and are eager to influence the rules based on community requirements/wants

Mike Digi

Pinnacle21

Response From ADaM team

FYI...

The check should add something like, “and xx>01”. That’s the conditionality bit. As it is, it isn’t valid to simply check for TRT01P and TRT01SDT --- it definitely is not an error if TR01SDT doesn’t exist and you only have TRT01P.

I remember a lot of discussion when we were finishing the IG about whether we need TRT01P for a single-period trial. It went back and forth, and ultimately ended up with, “it may not be useful but we’ll require it for consistency.” We obviously didn’t consider the consistency argument with regard to TRT01SDT/ TRT01EDT. We probably should have but there you go. It’s not in the book so it isn’t a rule.

Randall Austin

GlaxoSmithKline

I didn't forget :)

Over 2 years later, and this check now works as you would expect:

AD0078: TRTxxP is present and TRxxSDT is not present for periods greater than 01

AD0079: TRTxxP is present and TRxxEDT is not present for periods greater than 01

It will be released in the next version of the OpenCDISC ADaM Rules (v1.3) which will be aligned to the CDISC Validation v1.2 document.    This should be coming soon.

Take care

Mike D.