Top 20 Validation Rule Failures (CBER)

From PhUSE Wiki
Jump to: navigation, search

Contents


Purpose

The purpose of this project is to evaluate the top 20 validation rule failures experienced at CBER. The goals is to determine the cause and coordinate with CDISC, OpenCDISC, and FDA to find and propose a solution. Typically a rule failure is due to any number of the following factors:

  • Bad or incorrectly implemented rule
  • Unclear or conflicting underlying standards
  • Lack of understanding

Upcoming Meetings

The Top 20 Validation Rule Failures Project team meets every 2 weeks on Monday 11 am - noon EST.
To join the team, please contact [Hany Aboutaleb] or [Max Kanevsky]

Rules Under Review

SDTM Compliance

Rule Problem Proposed Solution Status Assignee Date Discussion
SD0002: Null value in variable marked as Required Under Review TBD Comment
SD0006: No baseline result in [Domain] for subject SDTM requires baseline for EG, LB, QS, VS, not all TEST have a baseline Validation rule and standard needs to align with use cases Under Review TBD Jul 30, 2012 Comment
SD0007: Inconsistent value for Standard Units Under Review TBD Comment
SD0026: Missing value for --ORRESU, when --ORRES is provided Not all tets have units
  • Identify tests with no units and update rule implementation
    • Smart check based on list in define or TS of tests that do not have units. Requires TS domain.
    • Alternative, controlled term in that field that indicates no units so don’t have to maintain table. This would impact listings
    • Add list of units and list of tests that have no units to define.xml
    • Add additional variable to the dataset that identifies the test as having no unit – NULL flavor. (this would be a record level not a test level soln.)
  • Propose solutions and determine pro/con of each before proceeding with solution.
Under Review TBD Aug 27, 2012 Comment
SD0029: Missing value for --STRESU, when --STRESC is provided
  • Not all tests have units
  • Sponsor doesn’t populate when there are units
  • Need to be able to differentiate between the two
  • Identify tests with no units and update rule implementation
    • Smart check based on list in define or TS of tests that do not have units. Requires TS domain.
    • Alternative, controlled term in that field that indicates no units so don’t have to maintain table. This would impact listings
    • Add list of units and list of tests that have no units to define.xml
    • Add additional variable to the dataset that identifies the test as having no unit – NULL flavor. (this would be a record level not a test level soln.)
  • Propose solutions and determine pro/con of each before proceeding with solution.
Under Review TBD Aug 13, 2012 Comment
SD0031: Missing values for --STDTC and --STRF, when --ENDTC or --ENRF is provided
  • Date collection issue.
  • Con meds for ongoing so have ENRF but no start information.
  • Business rule in IG/FDA explaining how to represent exposures or other events that happen in moments of time.
  • Controlled terminology to represent concepts such as started before trial and is ongoing with no known end date. Cant use this for the date but you need to be able to represent in some variable and then cross check against date field.
  • Check HL7 for this concept to see how they represent or what terminology is used.
  • --STRTPT could be used to represent that the event/intervention occurred prior to. If this solution is selected it would require clarification in the IG
Under Review TBD Aug 13, 2012 Comment
SD0035: Missing value for --DOSU, when --DOSE, --DOSTXT or --DOSTOT is provided Under Review TBD Comment
SD0063: SDTM/dataset variable label mismatch Under Review TBD Comment
SD0070: No Exposure record found for subject
  • Follow on studies, exposure occurred in first study
  • Randomized not treated (high risk trials – subjects die before treatment)
  • Not identified as screen failure
  • Phase 0 study looking at population, no exposure (gene/biomarker), used to select population for exposure
Need to clarify rule Under Review TBD Comment
SD0080: AE start date is after the latest Disposition date
  • May be a partial date issue
  • No disposition date or date is missing
  • May need business rule for example: Followed 30 days post exposure

DOES THIS CHECK PROVIDE VALUE TO FDA or SPONSORS?

  • Need disposition date for last contact so that there are no observations after the latest disposition, also look at disposition code list. Dates may be currently captured in ADaM as imputed dates. (FDA Direction)
  • Last pre-specified contact versus last contact impacts the way Sponsors define disposition events.
  • This check will fire for all date/time in study, if there is a start and end date for an observation the ck will fire twice.
  • Potential solution is to check against Subject Visit (SV) rather than as currently written
  • Alternatively only fire check when it is determined that the subject has left the study.
  • May need a date for last actual contact date.
Under Review TBD Aug 27, 2012 Comment
SD0081: Observation date is after the latest Disposition date
  • Need disposition date for last contact so that there are no observations after the latest disposition, also look at disposition code list. Dates may be currently captured in ADaM as imputed dates. (FDA Direction)
  • This check will fire for all date/time in study, if there is a start and end date for an observation the ck will fire twice.
  • Potential solution is to check against Subject Visit (SV) rather than as currently written
  • Alternatively only fire check when it is determined that the subject has left the study.
Under Review TBD Aug 27, 2012 Comment
SD0087: RFSTDTC is not provided for a randomized subject Under Review TBD Comment
SD0088: RFENDTC is not provided for a randomized subject
  • Not assigned and screen failure will not fire check.
  • In injectable studies this may not be populated.
  • Clarity on definition of reference start and end date.
  • May need additional check for new variables related to exposure start and end date.
  • What are Sponsors using this date represent?
  • Do all variables have to apply to all studies or is it possible that study design dictates where checks are applicable?
Under Review TBD Aug 27, 2012 Comment

Terminology Compliance

Rule Problem Proposed Solution Status Assignee Date Discussion
CT0027: Value for --OUT not found in (OUT) CT codelist Under Review TBD Comment
CT0037: Value for --BODSYS not found in (SOC) CT codelist Case sensitivity mis-match between CDISC Terminology and MedDRA
  • Confirm CDISC deleted codelist and OpenCDISC will have to rely on MedDRA codelist
  • OpenCDISC will need to removed this Check and activate new check
  • What is the requirement in SDTM for SOC regarding MedDRA versus other dictionaries?
  • Need to tie pieces together
Under Review TBD Aug 13, 2012 Comment

Define.xml Compliance

Rule Problem Proposed Solution Status Assignee Date Discussion
DD0010: ODM attribute should not be included in Define.xml Likely a lack of understanding by the users Need to clarify CDISC documentation and rule Under Review TBD Jul 30, 2012 Comment
DD0024: Invalid Codelist for variable
  • Non-extensible codelists in legacy studies
  • Extensible codelists need to be extended to include all terms utilized in the study in the define file
Suggest metadata guidelines for v2.0 to clarify extensible versus non-extensible codelist documentation Under Review TBD Aug 27, 2012 Comment

Other

Rule Problem Proposed Solution Status Assignee Date Discussion
SKIP_xxnnnn: SKIP messages Indicate that a domain (ex: DM) is missing and cannot be used for cross-dataset validation This is extraneous information and should be hidden from the user Under Review OpenCDISC Jul 30, 2012 N/A


Meeting Minutes

13-Aug-2012
27-Aug-2012


Back to Working Group 1: Data Validation and Quality Assessment Homepage


Last revision by DanBoisvert on 05/30/2013

Nobody voted on this yet
 You need to enable JavaScript to vote
Personal tools
Namespaces

Variants
Actions
Navigation
Projects
PhUSE Wiki Help
Toolbox